Marvin Ellison didn’t wait to be asked.

On Lowe’s Q2 earnings call this morning, the CEO volunteered, unprompted, that weather was the single biggest driver of his company’s regional performance this quarter — not tariffs, not rates, not the competitive environment he spent much of the rest of the call discussing.

“There is no material difference in geographies other than weather impacts,” he said. “Weather continues to be the biggest driver in geographic performance.”

That’s a notable thing for a home improvement CEO to say plainly, unprompted, in 2026. It’s also consistent with what he said a year ago, when weather got referenced 15 times across Lowe’s Q2 2025 call — by a wide margin the most weather-saturated retail call of last season.

Home Depot released results Tuesday, Lowe’s this morning.

Home Depot beat on both top and bottom lines and held its full-year guidance steady. Lowe’s beat on EPS but missed on revenue, and actually cut its full-year outlook — trimming sales, comp, and EPS guidance all down to the low end of its prior ranges.

But the two companies could not have told a more different story about how much weather mattered to get there. Here’s how the weather topic played out for both, compared to what they said last year.

Lowe’s

Ellison’s comment above isn’t a one-off. The pattern holds across both years.

Last year (Q2 2025): comps ran May -1.0%, June +0.3%, then surged +4.7% in July as weather “normalized” and outdoor categories caught up. Weather was referenced 15 times across the call — by my count, the most weather-saturated retail earnings call of last season.

This year (Q2 2026): comps ran May -0.4%, June +1.7%, July -1.2% — though that July number is distorted by the July 4th holiday landing a week earlier this year, which shifted roughly 75 basis points of comp from July into June on the calendar alone. Strip that out and the underlying trend was still soft. Total comp for the quarter: just +0.2%. Revenue came in at $25.96 billion against a Street consensus of $26.16 billion — a real miss — while adjusted EPS of $4.40 beat the $4.22 consensus, helped by an 11-cent tariff-refund benefit.

Weather wasn’t a footnote here either. CEO Marvin Ellison, unprompted, in response to a general question about regional performance, said flatly: “there is no material difference in geographies other than weather impacts... weather continues to be the biggest driver in geographic performance.”

He cited drought conditions in parts of the country and what he called “traumatic conditions in the Midwest.”

CFO Brandon Sink went further, tying the quarter’s -2.1% transaction decline directly to “pressure from weather-sensitive outdoor and seasonal categories.”

And when analyst Chris Horvers pushed on the year-over-year comparison, Sink pointed to a single event as the real driver: a rain-soaked Memorial Day weekend — “one of our biggest DIY events of the year... it kind of dragged on the full quarter results.”

Lowe’s didn’t put a dollar figure on any of this — the disclosures were all directional or in basis points, never in dollars.

But the setup was flagged ahead of time: Piper Sandler’s Peter Keith trimmed his Q2 comp estimate before the print, citing drought conditions and reasoning (correctly) that dry weather discourages exterior projects and softens demand for lawn and garden staples like plants, grass seed, soil, and fertilizer.

Total weather mentions this year: roughly 8, spread across the CEO, CFO, and two separate analyst questions, in both prepared remarks and Q&A. Fewer than last year’s 15, but still substantially more than Home Depot’s 5 — and unlike Home Depot, where weather only surfaced because an analyst asked, Lowe’s CEO brought it up on his own.

Home Depot

The contrast between this year and last year is stark, and it’s less about the weather itself than about how loudly management chose to talk about it.

Last year (Q2 2025), weather was the headline. CEO Ted Decker volunteered it unprompted: “weather did have a big impact.” U.S. comps ran +0.3% in May, +0.5% in June, then snapped to +3.3% in July as a heat wave broke across the North. Two separate analysts — Fadem and Gutman — pressed further on it. Seven total mentions of “weather” across the call, coming from both sides of the table.

This year (Q2 2026), weather barely got airtime. U.S. comps ran +0.5% in May, +1.2% in June, +2.2% in July — still an upward slope, but a far gentler one than last year’s dramatic swing.

Nobody on management brought up weather in prepared remarks. It only came up because analyst Chris Horvers asked directly, and when EVP of Merchandising Billy Bastek answered, he minimized it: “there’s actually just two really small pieces that drove that” — a calendar quirk (the year’s one heat week landing in the July reporting period instead of June’s) and one bad-weather week in May that “normalized” once excluded.

Five total mentions, all clustered in that single exchange, none of them CEO-level.

Worth noting: Ted Decker, Home Depot CEO, wasn’t on this year’s call, and Home Depot’s larger Pro and contractor mix gives it more insulation from DIY-driven weather swings than Lowe’s has — both of which are structural reasons this year’s call may have run quieter on weather, independent of whether the weather itself mattered less.

The analysis and writing here are mine. I use Claude as an editor — for fact-checking and line edits — not as a source of ideas or content.

© 2026 G2 Weather Intelligence. All rights reserved.

The frog never feels the doubling. It only feels today.

In 2007, I took a huge leap of faith, left my job, and joined an early-stage, venture-backed weather risk start-up in New York City called Storm Exchange. I was hired as Chief Strategy Officer, along with my (now long-time) pal Jarvis Cromwell, who came in as Chief Marketing Officer.

We had very strong early traction; our CEO/founder was very good at fundraising, and we closed a $7M Series B soon after Jarvis and I joined. We were on a roll and moved into nice office space in midtown Manhattan, hired a great team, and built a talented, growing development team in, of all places, Kyiv.

And then came 2008 and the great financial crisis. Our strong pipeline and early partnerships followed the stock market down; our investors pulled out, and we were forced to shut the doors.

For the first time in my career (since joining the US Air Force at 17), I didn’t have a job … this is when G2 Weather Intelligence first took shape.

I started consulting on weather strategy, and that work put me on stages and in front of rooms around the country, talking about climate and weather risk and the critical need to purposely plan for what was coming.

It was my argument for action …

Almost twenty years later, I was digging through old files for something else entirely, and found this presentation from that time … 20 years is a long time to sit in water that's been quietly getting warmer.

I presented this version at St. Mark’s Episcopal Church, in New Canaan, Connecticut. Not a boardroom. Jarvis’s church. As I recall, the rector was Reverend Peter Walsh (no known relation, but you never know).

My message then (as now) wasn’t built to sell anyone on weather data. It was meant to make a case for thinking differently about weather and climate risk, and the business case followed from that, not the other way around.

The presentation opened with my then mission statement, which is largely the same today: helping businesses and investors increase profits by building greater resilience in the face of increasingly volatile weather and climate … rather than reacting after the fact.

It included a photo of Eisenhower speaking to 101st Airborne “Screaming Eagle” paratroopers just prior to the D-Day invasion. Not subtle, but the point holds: the invasion was timed based on a weather forecast, not a hunch. Strategy built ahead of the weather beats heroics improvised inside it.

The photo has a personal edge, too. I was chief of weather operations for the 101st during Operation Desert Storm. You can read that story here —

The Wager Itself

The core of my talk borrowed from Blaise Pascal, who argued that even though God’s existence can’t be settled by reason alone, a rational person should live as though it’s true — because the believer who’s wrong loses nothing, and the skeptic who’s wrong loses everything.

Tim O’Reilly had made the case a year or two earlier that even if catastrophic warming turns out not to happen, the steps taken to avert it are still worthwhile — cleaner energy, new industry, less exposure to foreign oil, fewer of the hidden costs pollution leaves off the books.

My message framed it as a straight comparison: what’s the greater risk, acting on climate change and being wrong, or doing nothing and being right?

Stated that way, it isn’t really a climate question. It’s a decision-theory question. And decision theory doesn’t require consensus on the underlying science to produce a clear answer.

Cope and Avoid, Anticipate and Exploit

This wasn't a business insight first. I learned it supporting warfighters in the Air Force and Army, where the stakes make the distinction obvious: "cope and avoid" gets people hurt. "Anticipate and exploit" — like the weather call behind D-Day — wins the day.

• Cope and Avoid: ignore the forecast. Absorb the losses. Blame the weather when the numbers miss. Keep running the business exactly as before.

• Anticipate and Exploit: build the forecast into the plan. Build probabilistic weather forecast scenarios into the plan. Minimize the downside. Take the upside the Cope and Avoid businesses leave on the table.

Eighteen Years Later, the Bill Comes Due

Which brings me to why I went looking for that presentation in the first place.

I’d just read a New York Times piece — “Heat, Fire, Smoke and Storms Are Wreaking Havoc on the Economy“ — and it reminded me of the message I’ve been preaching for two decades. In fact, it reads like the invoice for the wager I described in 2009.

• A Chicago bakery owner loses 500 ice cream bars and nearly $30,000 in cakes to a storm-driven power outage, saved only by $1,700 of dry ice and a frantic round of calls to every favor she’d ever banked.

• A Los Angeles day care operator watches her electricity bill double from heat and smoke, some days deciding hour by hour whether the toddlers stay inside.

• Allianz Research estimates the most exposed developed economies could run 5 to 7 percent below their growth trajectory by 2030 if the next five years look like the hottest five of the last decade.

None of this is a new argument. It's the old argument, compounding.

Most businesses don't lack the ability to predict the weather. They lack the system to act on it — one that turns a forecast into a decision before the loss happens, not a phone call after.

What I Got Wrong, and Why It Doesn’t Undercut the Point

One thing I haven't mentioned about that talk: certainty wasn't the argument.

I wasn’t preaching inevitability. I was making a probabilistic argument — trust the signal, verify against the data, hedge accordingly. That’s the part I’d underline hardest today.

The wager was never "climate catastrophe is certain, so act." It was that acting makes sense whether or not you're convinced — because the cost of being wrong isn't the same in both directions.

That distinction is the whole discipline this newsletter is built on now. Weather and climate as a probability to be managed, not a belief to be argued.

Government Won’t Save Us

Here’s the part I’d say more bluntly now than I did from a church podium in 2009: waiting on government to manage this risk for you isn’t a plan (to put it mildly).

Public policy runs on electoral cycles measured in years. Weather risk runs on planning cycles measured in weeks and seasons. Those clocks don’t synchronize, and waiting for them to align is itself a decision, usually the expensive one.

That mismatch isn’t closing anytime soon. If anything, current policy is trending away from resilience, not toward it.

The wager was always a private-sector bet, because it had to be. It still is.

The Frog Doesn’t Know

The frog part of the metaphor is fiction. The part where nobody notices until it’s too late isn’t.

The temperature moves slower than the attention span of the business trying to track it. Nobody experiences 2009 turning into 2026 as a single dramatic jump. They experience it as one summer, then another, then a bakery owner explaining to a reporter that cake isn’t built for extreme weather.

When I gave that talk, the planet was running about a degree Fahrenheit warmer than normal. It’s now running over two degrees warmer. More than double, in the time it took a slide deck to become an archive file.

The frog never feels the doubling. It only feels today.

The analysis and writing here are mine. I use Claude as an editor — for fact-checking and line edits — not as a source of ideas or content.

© 2026 G2 Weather Intelligence. All rights reserved.

I’m resharing a post that I wrote last June on the coming disruption of the weather app market.

Hated by purist weather enthusiasts but increasingly loved by “old media” providers as a way to find new revenue sources (looking at you, CNN), there’s been a proliferation of plain-vanilla weather apps on the market.

Here’s the truth: Anyone (and their brother) with a halfway decent laptop, an internet connection, and a subscription to Claude or ChatGPT can build a workable weather app. And it feels like they are, one “AI-powered forecast” launch at a time.

Still, I stand by my prediction. Weather apps won’t go away completely, of course, but the likes of weather.com and AccuWeather and, yes, CNN Weather will find themselves up against smarter, more tailored, and more activity-specific (I’m new to it but a big fan of OpenSnow … and I don’t even ski) applications that directly influence decision-making.

That shift — from information to decision-making — is exactly what I was getting at last summer, so I thought I’d share those thoughts again.

Here's the original post, from June 2025:

A recent deep dive into the next phase of artificial intelligence makes one thing brutally clear: we are on the verge of a technological shake-up as significant as the launch of the App Store itself. And it raises a critical question for a category that hasn't fundamentally changed in a decade:

Does this signal the end of the weather app as we know it?

The race to build useful AI agents isn't about making chatbots that can write a better poem. It's about creating autonomous "doers" that can perform complex, multi-step tasks in the real world on your behalf.

While we've all been focused on the flashy Large Language Models (LLMs), the real work is happening in the background, building the "plumbing" that connects these AIs to the tools and data they need actually to accomplish tasks.

And this is where your favorite weather app faces an existential crisis.

Currently, your weather app serves as an information provider. It's a digital window you look through. It might be a very shiny, well-designed window with beautiful radar maps, but all it does is show you what's outside.

The coming wave of AI agents won't just show you the window—they'll close it for you when it starts to rain, turn up your thermostat, and text your partner to remind them to grab an umbrella.

The key challenges the AI industry is solving right now are the very things that will transform weather services from passive apps into proactive concierges:

1. Connecting to Tools: An agent's true power comes from its ability to use other apps and APIs. This is the foundation for a weather service to finally break out of its data silo.

2. Gaining Permission: This is the game-changer. For an agent to truly serve you, it needs your permission and authorization to securely access your other data—your calendar, your flight details, your transit app, your health data. New standards and technologies are being built right now to solve this complex problem.

Think about the difference.

• Today's Weather App: "There is an 80% chance of rain at 2 PM."

• Tomorrow's Weather Agent: "I see you have 'Park Picnic with Sarah' on your calendar for 2 PM. There is now an 80% chance of rain starting at 1:30 PM. I've found three alternative indoor cafes nearby with great reviews that are available. Would you like me to reserve a table and text Sarah with the updated plan?"

It's about synthesizing information from your email, real-time flight tracking, and your calendar to deliver a complex, actionable answer.

That's the future. A weather concierge that doesn't just give you a forecast but integrates it with your life. It will know your flight is delayed due to a storm and can re-book your Uber. It will see that you have an outdoor run scheduled and, based on your health data and the air quality forecast, suggest a better time or an indoor alternative.

So yes, this is the beginning of the end for the weather app as we know it.

The era of single-purpose information apps is drawing to a close. The future isn't another beautifully designed radar map; it's a trusted, permissioned AI agent that manages the weather's impact on your life so you don't have to.

The question for every weather app developer today is simple: are you building a better window, or are you building the intelligent agent that will tell us when to close it?

Musical Coda

Big news today … and I’m proud to be part of it.

ERaaS Health builds its climate resilience platform — the member identification, the environmental intelligence, the risk targeting — on top of Hippocratic AI’s clinical-grade voice technology.

It’s a genuine partnership: Hippocratic AI’s platform makes the rapid-response calls possible at scale, and ERaaS Health’s environmental intelligence determines who needs to be reached, and when.

EmblemHealth’s results, cited below, are what that partnership looks like in the field: over 50,000 heat outreaches ahead of NYC’s July 4th heat dome, tens of thousands more across cooling-benefit and wildfire-smoke campaigns, and a 9.02/10 member satisfaction score along the way.

Full release below, verbatim.

Hippocratic AI Announces New Suite of Rapid Response Climate Agents as Extreme Weather and Record Breaking Heat Escalate Health Threats

Already Proven: Hippocratic AI and EmblemHealth Demonstrate How Rapid-Response Climate Resilience Agents Are Helping Vulnerable Members. Expanded Solution Now Available for Payers and Providers

MENLO PARK, Calif.--(BUSINESS WIRE)--Following record-breaking temperatures that affected more than 200 million Americans over the Fourth of July weekend and on the heels of back-to-back heat waves that have set national records across much of Europe, Hippocratic AI today announced that its climate-responsive voice AI agents are now available for health systems and payers across the country to deploy immediately, as healthcare organizations prepare for ongoing heatwaves and extreme weather brought on by El Niño.

Extreme weather, smoke, allergens: Conversational AI is proving successful at reaching, aiding vulnerable members.

The launch of rapid-response resilience agents builds on the recent successful outreach conducted by EmblemHealth and other clients who have deployed the agents to help vulnerable patients during extreme heat, cold, smoke and other factors.

The outcomes are real: During one recent heat wave, EmblemHealth’s bilingual voice AI agent, Rachel, reached more than 16,000 members in 24 hours, identified that 10% needed further help, and earned a 9.02 out of 10 member satisfaction score. The agent, powered by Hippocratic AI, conducted clinical heat stroke assessments, safely routed vulnerable individuals to local cooling centers, and identified urgent health issues. Since then, the program has expanded to over 50,000 heat outreaches to prepare members in advance of the recent July 4th weekend heat dome in NYC, a set of 15,000 cooling-benefit outreaches conducted in April and May — and most recently calls to prepare vulnerable members from exacerbations due to bad air quality in NYC driven by the Canadian wildfires.

The agents, built on the only generative voice AI platform proven safe enough for clinical interactions, can check on vulnerable patients, provide valuable education, assist in getting transportation to cooling centers and escalate when a member is in danger. They have already been proven in the field on both ends of the temperature scale: reaching members during New York’s last 2 summers of historic heat waves, and standing ready for cold-weather events, wildfire smoke, and seasonal flu surges.

EmblemHealth is able to identify its most vulnerable members using risk factors such as the urban heat island effect, heat vulnerability, social vulnerability, clinical risk, and medications, informed by precision environmental intelligence, to target outreach to members most at-risk of climate-related illness.

Extreme heat and cold, air-quality risks, and other extreme weather incidents can aggravate asthma, COPD, cardiovascular disease, and diabetes — the chronic conditions that already drive significant U.S. healthcare spending and put millions of Americans at risk. Most care teams cannot reach their highest-risk populations at the scale or speed a weather emergency demands. Hippocratic AI’s voice agents close that gap with high-volume concurrent call capacity, clinical-grade safety, and the conversational empathy needed to keep people engaged, especially those who are older, isolated, or managing multiple conditions.

Forecasters at the World Meteorological Organization and NOAA expect El Niño to strengthen rapidly through the fall, raising the odds of more frequent and more severe heat waves, droughts, and extreme weather worldwide, making durable, always-on outreach infrastructure a necessity rather than a convenience for the populations most at risk. Historically, capacity, not intent, has been the binding constraint on reaching vulnerable populations during these events. Hippocratic AI’s voice agents extend the human workforce rather than replace it, and are built to scale instantly when an alert is issued, whether that alert is a heat advisory, a cold-weather warning, or a wildfire smoke event.

Built for every climate event, not just one

The same agent infrastructure that responded to the Northeast heat emergency already extends across a full range of climate and health events:

• Allergen and air-quality alerts: agents proactively reach members with asthma, COPD, and seasonal allergies when pollen counts spike or air quality deteriorates, confirming inhaler access, refill status, and trigger-avoidance plans.

• Year-round climate-health response: from wildfire smoke to extreme cold to flu surges, health systems and payers can activate a Resilience Agent the moment a region issues an alert, reaching members who are not being served today.

• Equity reach: the same infrastructure extends to rural members, Spanish speakers, and high-utilization patients who have historically been underserved by traditional outreach.

© 2026 G2 Weather Intelligence. All rights reserved.

It’s been refreshingly cool in the northeast over the last week or so. So cool that I’ve worn long sleeves on a couple of my morning walks with Benji.

It reminded me of this post from approximately the same time last year. Late June and especially July were brutally hot, but when we moved into August, it turned refreshingly cool and served as a harbinger for very early fall and back-to-school demand.

You can read my take from early August last year here:

Over the last week or so here in the western Philly suburbs, the weather has been absolutely beautiful, with low dew points, cool mornings (high 50s, anyone?), and clear blue skies. Literally sweatshirt weather on our morning walks at Struble Trail.

But there’s a difference this year from my bullish take last year on early fall demand—this year’s cool weather in the east is a head fake. Or at least that’s what the National Weather Service says.

Last August brought relatively normal temperatures in the Northeast (which, given our warming climate and the earlier heat, made it feel “fallish”) and near-record cooler-than-normal temperatures in the Southeast.

This year, there is more heat to come in August, and the probabilities of hotter-than-normal and (much!) hotter-than-last-year temps across the major population centers in the Southeast and Northeast are essentially locked in. You can take it, literally, to the bank

Here’s last year’s ranking from normal …

And this year’s temperature forecast probabilities …

Why the break itself is the risk

There’s a well-established finding in heat epidemiology that the first heat wave of a summer is measurably more dangerous than the ones that follow it. A 43-community U.S. study found mortality rose 5.04% during a summer’s first heat wave, compared with 2.65% during later ones — roughly double the effect, from the exact same temperatures.

The explanation is physiological. Heat tolerance builds with exposure: sweat response improves, cardiac output adjusts.

However, that adaptation doesn’t last. Research on the mechanism describes it decaying within days to weeks once the heat stops.

That’s not a footnote here. It’s close to a description of the exact stretch the East just lived through. A week of relief sits comfortably inside the window where acclimatization built up over three weeks of heat and smoke can meaningfully lapse.

The population heading into August’s forecasted heat hasn’t been resting. It’s been quietly losing ground.

The baseline trap hiding in plain sight

Here’s where this becomes a planning problem, not just a health one.

Go back to last August across the East. The Southeast’s regional temperature rank for August 2025 came in at 8 out of 131 years on record — near-record coldest — a sharp reversal from a record-setting heat dome that hit the East in late June 2025, one large enough that NOAA counted more than 100 million people across 726 counties experiencing record heat in a single stretch.

The Northeast didn’t see the same dramatic swing, but it told the same story in a lower key — an unremarkable, near-average August, ranking 87 out of 131, the kind of month that never registers as an anomaly on anyone’s radar.

Here’s the message underneath both of those numbers: a cold August in the Southeast, especially one that close to a record, is now a genuinely unusual event, not a routine one.

Record daily highs have been outpacing record daily lows nationally by roughly 2 to 1 over the past two decades, and researchers project that ratio could widen to as much as 15 to 1 by the end of the century at the current pace of warming.

In a stable climate, that ratio sits near even. It doesn’t anymore. A month that lands near the cold end of 131 years of records isn’t the new normal reasserting itself — it’s the outlier, made rarer every year by the same warming trend that makes this year’s above-normal outlook the more statistically expected outcome.

Put plainly: last August wasn’t a preview of a cooling trend. It was a low-probability event, in a climate that’s making low-probability cold events rarer every year, sitting right next to a high-probability warm one this year.

Any payer or health system pulling last August as a planning baseline is comparing against an anomaly and calling it a baseline.

Given the odds — above-normal heat forecast against a near-record-cool comparison year — an increase in emergency room visits this August isn’t a surprise to be explained after the fact. It’s the expected outcome.

The same trap, one aisle over

It isn’t just health systems reading last August’s numbers. It’s retail.

Cooling equipment — window units, portable AC, fans — is exactly the kind of weather-sensitive category this newsletter has spent all year tracking on earnings calls.

If large stretches of the East came in near-record cold or just quietly average last August, it’s a reasonable bet that cooling-equipment sell-through across the region came in soft right alongside it, the same demand collapse playing out in a different aisle. The retail version of this trap is arguably worse, because a retailer can’t fix it in real time.

A healthcare payer can redirect outreach within days using technologies like ERaaS Health's (I'm an advisor), which moved from data to member outreach in under 24 hours during last summer's heat events.

A retailer's regional inventory position for August was locked in months ago, against a sales plan built substantially on trailing comps.

If last year’s soft Eastern numbers shaped this year’s allocation or reorder points, and the region snaps back to above-normal heat as forecast, the lead time on a window AC unit doesn’t bend to match a forecast that changed after the purchase order was already cut.

That’s how a soft sales miss becomes an actual stockout, timed to arrive exactly when demand spikes.

Same mistake, same population, twice

Both traps come from the identical error: reading a mild year as the trend instead of the anomaly. And they don’t stay separate. A stockout on window units isn’t just a missed sale. It’s a household without functioning cooling during a heat event that historical comps said wouldn’t be unusual — the same population a health system’s (potentially) under-staffed outreach program was also least prepared to reach.

The huge risk is that last year was (relatively) cool, and cool is getting statistically rarer every year the climate keeps warming. Anyone planning around last August is planning around the exception, not the rule.

Musical Coda

© 2026 G2 Weather Intelligence. All rights reserved.

I’m pretty active on LinkedIn. Maybe you’ve noticed.

And, like me, you may have noticed a disturbing trend over the past year or so: a shitstorm of AI-generated dreck produced by “thought leaders” who are prompt-engineering thinly (and sometimes not so thinly) disguised sales pitches to flog products and, hopefully, generate leads.

What bothers me most is that some of it (more than I’d like) reads like posts I’ve written on Substack, with help from Claude and friends.

That’s right, I feel seen.

It’s been nagging at me for a while, to the point where I’ve questioned my own approach, and even whether it’s worth continuing to write in a medium that’s getting hollowed out in real time.

So I was encouraged, honestly, by Substack’s announcement this week about putting AI detectors — aka BS detectors — into the app.

In their own words:

I've been writing this newsletter for about two years now, with AI in the loop most of the time, and I've learned — and am still learning — how to use the technology without handing it the keys to vomit out a newsletter under the G2 Weather Intelligence name.

The Wand Doesn't Make the Magician

An AI detection score, whatever number it lands on, is asking the wrong thing. The real question isn’t whether AI touched the words. It’s where the thesis underneath them actually came from.

AI can produce a fluent, confident, well-organized paragraph about almost any subject on earth. What it doesn’t have is thirty-plus years of sitting across the table from retail, CPG, and healthcare decision-makers, learning which weather signals actually move a P&L and which are just noise, where the real complexity lives in turning a forecast into a staffing call or a markdown decision, and how much upside is sitting there unclaimed for the organizations willing to act on it.

Take that away, and what comes back isn’t insight. It’s algorithmic hand-waving — polished, plausible … and empty.

How I Use AI to Write G2 Weather Intelligence

AI knows a lot, but it doesn’t (yet) know what I know. What’s in my head is what I’m putting into these dispatches — accumulated from working with senior leaders in the retail, CPG, and healthcare industries, and before that, warfighters in the Army and Air Force.

Here’s the process I’ve developed since starting the G2 Weather newsletter —

1. Start with the thesis. The points I want to make are always my own, written or dictated by me first.

2. Get an outline, and push back on it hard. An outline that survives untouched usually means I didn’t push hard enough.

3. Get a first draft.

4. Edit it by hand, section by section. Rewrite what doesn’t sound like me, cut what does but doesn’t belong. Maybe add a Walsh-ism or two. Maybe even a non-gratuitous four-letter word.

5. Bring AI back in for line edits and fact-checking. Every claim gets checked against a real source before it goes anywhere near “publish.” That step isn’t decorative — it’s caught a retracted study and a misattributed statistic in recent drafts, the kind of errors that would have quietly cost this newsletter the credibility it runs on.

6. Final review, start to finish, to make sure the piece holds together as one argument instead of a stitched-together set of good paragraphs.

7. Add a musical coda (h/t ), if it’s the right kind of post.

8. Publish.

At no point in that sequence does an AI decide what I believe, what I’m willing to put my name behind, or what gets cut because it’s wrong. That part doesn’t scale, and it isn’t supposed to.

Force Multiplier, Not Author

I’m a raging advocate for AI, not a skeptic of it.

G2 Weather Intelligence exists to make the case that these tools, used systematically and with discipline, are among the most critical instruments we have for building a more resilient society — whether that’s a health plan reaching a member before a heat wave, a retailer adjusting staffing ahead of a snowstorm, or a newsletter that fact-checks itself before it publishes.

None of that argues for less AI. It argues for more of it, applied by people willing to be accountable for what it produces.

That, my friends, is AI’s superpower.

Musical Coda

© 2026 G2 Weather Intelligence. All rights reserved.

After IBM and The Weather Company, I joined BreezoMeter as General Manager for North America in late 2021 — helping bring what I believed was the world’s most precise air quality intelligence platform to the US market.

We were out front on this story.

I spent much of my time at BreezoMeter making the case publicly — in media appearances, industry conferences, and briefings with healthcare and insurance executives — that air quality was a business intelligence problem, not just a public health advisory. The data existed. The forecast capability existed. What was missing was the organizational infrastructure to act on it before the emergency room filled up.

That argument felt urgent in late 2021. It feels existential in July 2026. A year after I joined, Google acquired BreezoMeter. The thesis was validated at the highest possible level.

Walking Benji this morning, the smoke was visible before we’d gone a block. You could smell it. The sun hung orange and small in a gray-brown sky. It felt like Mars.

Déjà vu. All over again. The Canadian wildfires are back, the smoke is here, and the forecast that made this entirely predictable was available 48 hours ago.

This Is Not a Surprise. It’s a Pattern.

The 2023 Canadian wildfire season burned 15 million hectares — more than doubling the previous record and turning New York City’s sky orange in an event that made international headlines.

In 2026, close to 40,000 fires have already torn through more than 3.6 million acres across the US — almost 10,000 more fires and nearly a million more acres burned than the 10-year average for mid-July. Roughly 850 fires are burning across Canada right now, including more than 100 in Ontario. An enormous heat dome over the upper Midwest has created a clockwise air rotation, scooping up smoke from Canada and Minnesota and carrying it eastward across the most densely populated corridor in the country.

The science of smoke transport is well established. When wildfires burn in specific Canadian provinces under specific wind patterns, the smoke follows predictable atmospheric pathways southward into the US. NOAA’s Air Quality Forecast Guidance and commercial air quality platforms can model that transport with meaningful accuracy 24-72 hours in advance.

The event is not unpredictable. The response should not be reactive.

The data on the trend is unambiguous. Per-person annual exposure to wildfire smoke in the US was four times higher on average between 2020 and 2024 than during the prior fourteen years, according to Stanford University’s Environmental Change and Human Outcomes Lab.

Climate change is the driver — research published in 2025 found that anthropogenic climate change contributed 33 to 82% of the observed increase in burned area in the western United States from 1992 to 2020, and accounts for nearly half of the increase in surface smoke PM2.5 concentrations.

This is not a weather story. It is a climate story. And it’s accelerating.

The Health Stakes Are Not Subtle

The academic literature on wildfire smoke and human health has been accumulating rapidly, and the findings are consistent. A systematic review and meta-analysis of over 124 million patients published in 2025 found that for every 10 μg/m3 increase in PM2.5 exposure, wildfire smoke drives a 4% increase in emergency department visits, a 4% increase in hospital admissions, and a 2% increase in all-cause mortality.

The pattern holds hardest at the edges of the population.

Pediatric asthma-related ED visits may increase by over 30% during wildfire events. Elderly COPD patients face heightened hospitalization and mortality risk. Between 2019 and 2024, wildfire smoke caused an estimated 24,000 to 62,000 asthma emergency department visits in the US alone.

The smoke blanketing the Eastern US today is not just an acute health event. It’s a chronic health risk that accumulates with every exposure.

The New York event during the June 2023 Canadian wildfire smoke episode remains the most documented US case study. A CDC field report analyzing all 134 emergency departments in New York State outside New York City found that asthma-associated ED visits increased markedly when air quality reached its poorest levels during the event.

What BreezoMeter Taught Me — and What Comes Next

When I was working to develop BreezoMeter’s North American business, the core insight was that air quality data was everywhere, but actionable air quality intelligence was almost nowhere to be found.

The same Color Code system that had been telling people “it’s a red air quality day, sensitive groups should limit outdoor activity” since the 1990s was still the primary public health communication tool for a risk that had grown four times more severe in the preceding decade.

Google bought BreezoMeter because they understood that environmental intelligence embedded in products people already use — Maps, Search, Android — reaches people at the moment of relevance rather than in a general advisory that most people ignore.

The next step in that thesis is what ERaaS Health and Hippocratic.ai are building — environmental intelligence embedded in health-intervention infrastructure rather than in consumer apps.

Not "it's a red air quality day" sent to everyone. But a message from a conversational (and multilingual) weather intelligence agent, in this case, Rachel, sent directly to a specific member with COPD in zip code 10025:

“The air in your neighborhood is expected to turn very unhealthy within 24 hours; here's what that means for you today; make sure your rescue inhaler is refilled”

That is not a public announcement. That is a specific health intervention—enabled by AI. The difference between those two things is measurable in fewer emergency room visits, fewer hospitalizations, and, ultimately, lives saved.

From Instinct to Infrastructure

BreezoMeter’s thesis was that air quality data should be as accessible and actionable as a weather forecast. Google paid for that thesis.

ERaaS Health and Hippocratic.ai are building the next one — health intelligence as proactive and precise as the environmental data that predicts it.

Wildfire smoke events affecting the Eastern US are no longer rare. They’re a predictable feature of every summer — more frequent, more intense, and more foreseeable than the public health response currently reflects.

Benji and I cut the walk short this morning. I closed the windows. I checked the AQI. Individual protective action is instinctive. Organizational proactive response — identifying at-risk members, initiating outreach, closing the gap before the ER fills up — requires infrastructure that matches the speed of the forecast.

That infrastructure exists. That’s the mission ERaaS Health and Hippocratic.ai are built around: actionable intelligence that saves lives.

Musical Coda

© 2026 G2 Weather Intelligence. All rights reserved.

We’ve been spending a lot of time at our cottage on the Delaware shore this summer — fishing, walking the dog, cutting grass, and generally enjoying the kind of schedule that comes with being editor-in-chief of your own independent publication.

Which is to say, I make my own schedule, and lately the schedule has called for more time with a rod in the sand and less time in front of a screen.

I'll be honest — I've thought about pulling an Irish goodbye on the Substack. Just quietly disappearing into the summer.

But somewhere along the way this publication found an audience of over 700 subscribers and followers who actually show up for it, and that feels like an obligation worth honoring even when the fishing is good, and the heat index in Philadelphia is making the Delaware shore look like the smartest real estate decision I ever made.

The Weather Didn’t Take a Vacation

Late June and July are when extreme heat has its biggest impact on consumer spending and, more importantly, on public health — and this summer delivered on both fronts while I was gone.

The weather didn’t take a vacation just because I did.

The July 4th heat dome delivered exactly what the forecast had been showing for more than a week.

Philadelphia’s Fourth of July parade was canceled. Washington’s Independence Day parade was canceled. Two of the most iconic American holiday traditions, in two of the most historically significant cities in the country, were shut down by the same weather system the models had been tracking since late June.

The National Mall fireworks went ahead anyway — 850,000 pyrotechnics over forty minutes, the largest display in history according to the organizers. Under the stagnant atmospheric conditions that had been baking the capital for days, the smoke had nowhere to go.

According to air quality data provided by Ambee — a climate intelligence platform that turns environmental data into actionable business intelligence — PM2.5 levels peaked at more than six times normal, crossed the EPA threshold for “unhealthy” air, and lingered for hours after the display ended, drifting as far as fifteen miles downwind into neighborhoods still recovering from the heat.

Airborne magnesium — used to create bright white flashes in fireworks and capable of irritating skin and lungs — increased by more than 9,000 percent.

Before the show, the National Park Service had conducted its own analysis projecting "hazardous" air quality near the National Mall during and after the display, and "very unhealthy" conditions in downtown Washington, Capitol Hill, and across the Potomac in Arlington.

That analysis was not released to the public — an omission that current and former Park Service employees called … unusual.

Heat Doesn’t Go to the Hospital. People Do.

Most people think about extreme heat the way a TV meteorologist presents it — it’s going to be hot, stay hydrated, find air conditioning. Healthcare systems think about it very differently, because the heat itself is never the patient.

The person suffering heat stroke is the patient. The person arriving dehydrated and in cardiac stress is the patient. The COPD patient whose symptoms worsened after breathing six times normal PM2.5 levels at 5 AM on July 5th is the patient. The elderly resident in a north Washington neighborhood who couldn’t get a wellness check in time is the patient.

The heat didn’t go to the hospital. The people the heat hurt did.

Weather changes how many people need care, what they present with, and when they arrive — in a sequence that the forecast makes visible well in advance of any of it happening.

That is not a meteorological observation. It is an operational one, and it changes everything about how a health plan, a hospital system, or an employer with outdoor workers should be thinking about the next heat event that is already showing up in this week’s forecast.

The Tools Being Built for This Moment

This is exactly the problem that ERaaS Health is built to address. I serve as an advisor to the company, founded by Samrat Kulkarni, which gives me a close view of what proactive weather-informed health intervention looks like in practice.

ERaaS provides the intelligence layer — identifying at-risk members before the event peaks, triggering proactive outreach, and surfacing escalations to health plan care teams for wellness checks, cooling center referrals, and emergency interventions.

The deployment layer is Hippocratic.ai — an AI agent platform purpose-built for healthcare outreach at scale. ERaaS’s weather-informed risk intelligence combined with Hippocratic.ai’s deployment infrastructure is what makes proactive intervention possible across thousands of members simultaneously, at the speed the forecast window requires rather than the speed of the emergency room intake queue.

During last summer’s heat events, the system demonstrated what that looks like in practice — moving from data to proactive intervention in under twenty-four hours, identifying at-risk members, initiating outreach through an AI agent, and surfacing escalations to care teams for targeted support.

The results among engaged members were striking: a 47% reduction in ER and inpatient admissions, nine out of ten member call satisfaction, and an estimated 5-to-1 return on investment.

Not a pilot. A deployed system. Running at the speed of the forecast.

While I Was Fishing, Rachel Was Working

I came back from the shore rested and slightly sunburned, with a healthy appreciation for what a near-normal temperature pattern feels like when the alternative is a 110-degree heat index ninety miles to the north — only to find another heat event already building across the Eastern US this week.

The atmosphere runs on its own schedule.

It doesn't care that organizations are still recovering from the July 4th heat event. The signal has been visible for more than a week — above-normal temperatures forecast across the Mid-Atlantic and Southeast, another round of elevated heat index values, another predictable sequence of health and demand impacts building in the pipeline.

The heat didn’t take a vacation. The signal didn’t either. While I was packing up the cottage, ERaaS Health’s AI agent Rachel was already hard at work — reaching out proactively to vulnerable members in New York City ahead of this week’s heat event. Not waiting for the emergency room to fill up. Acting on the forecast.

Weather is a predictable driver of population health risk, yet most healthcare organizations still aren’t acting on it proactively at scale. ERaaS and Hippocratic.ai are built to change that.

Actionable intelligence that saves lives. That’s the mission we share.

© 2026 G2 Weather Intelligence. All rights reserved.

I spent five years at IBM Consulting after the company acquired The Weather Company and saw firsthand what happened when a retailer or CPG company sought to build an integrated weather strategy.

The business case was clear. The data was available — and relatively inexpensive. The financial opportunity was real and measurable. But the consulting financial model made the math not work.

Billable hours. Utilization targets. Multi-million dollar engagement economics. A large team of consultants required to connect weather intelligence across seasonal planning, demand forecasting, pricing, marketing, staffing, and financial risk management. A multi-year transformation timeline before the ROI showed up in the P&L.

Most retailers looked at that equation and made a rational decision: pass. Or pilot. A demand forecasting experiment here. A weather-triggered marketing campaign there. Point solutions that never connect to the integrated system that delivers real value.

Weather intelligence stayed tactical. Not because the opportunity wasn’t real. Because the consulting model made it too expensive to justify.

The WSJ Story This Morning

This morning, the Wall Street Journal reported that consulting firms are in what one industry observer called “an existential scramble” to find a new pricing model as AI upends traditional revenue streams.

A Deloitte executive told consultants at a town hall that traditional labor-based hourly work is expected to shrink to a sliver of the total market by 2035. AI agents are expected to grow exponentially to become a majority of the professional services market.

One Deloitte consultant’s takeaway: “They heavily implied our model is toast.”

Pat Petitti, CEO of AI-based consulting platform Catalant, was more direct: “AI is destroying their business model.”

While this represents a financial risk for the consulting industry, it is a structural opportunity for retailers, CPG companies, and healthcare organizations with weather-sensitive operations — to build an integrated weather strategy that the old consulting model made too expensive to justify, at a cost and speed previously unimaginable.

What AI Agents Change

AI agents change the value equation in two fundamental ways.

First, they dramatically reduce the human labor required to build and operate the integration layer. The work that previously required a large consulting team — correlating historical weather data against point-of-sale data by category, geography, and time window; building triggering mechanisms; connecting the forecast to the operational response across every enterprise function — can now be done by AI agents at a fraction of the cost and in a fraction of the time.

Second, they make outcome-based pricing — which the WSJ identifies as consulting’s only viable future — actually work for weather intelligence engagements. When the system can demonstrate a measurable reduction in markdown depth, a measurable improvement in full-price sell-through, or a measurable lift in weather-driven category revenue, the value is quantifiable before the engagement ends. That changes the risk calculus for the buyer entirely.

As noted, that shift represents a financial risk for the large consulting firms built around the old model — and a significant opportunity for the consumer-facing organizations they have historically served.

Retailers, CPG companies, and healthcare organizations can now access the same integrated weather intelligence capability that previously required a multi-million-dollar engagement with a handful of strategy consultants or a small internal staff, at a fraction of the cost and timeline.

The barrier was never the data. The barrier was never the technology. The barrier was the cost and complexity of the integration layer that connected them.

AI agents are removing that barrier—and it’s happening now.

Great News for Retailers, CPGs, and Healthcare. Not Great News for Large Consulting Firms.

The companies that move first — with a small team, a clear framework, and the right data — will build a capability that their competitors can’t easily replicate. Not because the weather data is proprietary or expensive. It isn’t. The data is table stakes.

The competitive advantage is the organizational integration — every function responding to the same weather signal simultaneously. That is genuinely hard to build. And genuinely valuable once it exists..

The large consulting firms that built their business around the old model face a structural challenge the WSJ describes clearly: “Many are being forced to cut prices before they themselves have actually realized the cost-saving gains from the technology.”

The transition is slow and painful for organizations built around utilization and billable hours.

For retailers and CPG companies, that transition is an opportunity. The expensive barrier is coming down. The playbook for building an integrated weather strategy without a multi-million dollar consulting engagement is available right now.

The same principle applies in healthcare — where weather intelligence is increasingly being used to identify at-risk populations before a heat emergency peaks rather than after the emergency room fills up. A separate but parallel opportunity, and one that is equally underserved by the old consulting model.

The Brainstorm Series Is That Playbook — For Retail and CPG

The G2 Weather Intelligence Brainstorm series is built specifically around retail and CPG — the categories where the weather-to-demand relationship is most documented, most measurable, and most directly connected to earnings performance.

Each post is a module in a framework designed for exactly this new model. Not a science experiment. Not a multi-year transformation program. A connected framework that a small, focused team can implement systematically.

Seasonal planning sets the buy. Demand forecasting optimizes the distribution. Pricing manages the margin. Marketing activates the demand. Staffing schedules for the forecast. Financial risk management closes the loop.

The capstone post will show what it looks like when it all runs simultaneously — in a fictional retailer set in the near future, looking back at how they built the capability that their competitors were still piloting.

The consulting model that kept weather strategy on the shelf is breaking down. The AI agents that replace it are being built right now. The framework is here. The barrier is gone.

It’s time to move on.

Musical Coda

© 2026 G2 Weather Intelligence. All rights reserved.

The forecast high at Citizens Bank Park: 104 degrees. (!)

We’ve already purchased hand fans and cooling towels and are hoping the forecast isn’t as dire as it looks. But I’ve spent more than thirty years in the weather forecasting business, and the signal has been unambiguous and persistent for the last week — it’s going to be a Fourth of July scorcher.

And this is not a one-day event. The 8-14 day outlook extends the above-normal signal through July 13th

The Health Story Comes First

A 104-degree day at a baseball stadium is uncomfortable for a healthy adult. For an elderly person without air conditioning, a very young child, or someone with a chronic health condition, it can be life-threatening.

Last summer’s heat events killed 19 New Yorkers in four days. Most died at home, without air conditioning. The NYC Health Department estimates approximately 500 New Yorkers die from heat-related causes every year — the vast majority quietly, without making the news cycle. NYC experienced four separate heat waves last summer, finishing 2025 as its second-hottest summer on record.

Western Europe is already there. Multiple countries issued red alerts — warnings of “a risk to life for even the healthy population” — as temperatures exceeded 104 degrees Fahrenheit across France, Germany, Switzerland, and Italy. French nuclear power plants cut output by approximately 7% of total demand as high temperatures limited access to cooling water. Rail networks, schools, and working hours were disrupted across the continent.

The same forecast system that is showing 104 degrees at Citizens Bank Park on July 2nd showed those European temperatures weeks in advance. The question is never whether the signal is visible. It is whether the systems are in place to act on it before the emergency room fills up.

AI-First Health Resilience — At the Speed of a Heat Wave

This is exactly the problem that ERaaS Health is built to address. I serve as an advisor to the company, founded by Samrat Kulkarni. ERaaS develops AI-first health intervention tools specifically targeting heat-related illness and mortality risk — working with health insurance companies to identify at-risk members before a heat emergency peaks and intervene proactively.

During last summer’s heat events, ERaaS demonstrated what prevention looks like at the speed of a forecast. The system moved from data to proactive intervention in under 24 hours — identifying at-risk members, initiating outreach through an AI agent, and surfacing escalations to health plan care teams for wellness checks, cooling center referrals, and emergency interventions.

The results among engaged members: a 47% reduction in ER and inpatient admissions. Nine out of ten member call satisfaction. A 5-to-1 estimated return on investment.

As heat events become more frequent and more severe, the ability to move at the speed of the forecast — not the speed of the emergency room — is what separates a preventable death from a statistic.

Investors Are Paying Attention

The financial markets are beginning to price what the forecast has been signaling for years.

CNBC reported last week that investors are actively rethinking portfolio positioning in response to the European heat wave. Stephanie Niven, co-portfolio manager of the Global Sustainable Equity strategy at Ninety One, told CNBC that her team views the rise in extreme weather as “a structural growth opportunity” — seeking companies with products and solutions that “lean into helping people respond and build resilience in a challenging time.”

Her fund’s holdings include insurers such as Aon and Intact Financial — companies that are building more sophisticated climate modeling into their risk systems. And cooling and refrigeration manufacturers like Trane Technologies, which makes the physical infrastructure that keeps people alive during extreme heat.

Morningstar’s chief equity strategist Michael Field identified Johnson Controls and Siemens as major beneficiaries — both operate in the HVAC space where commercial heat pumps that double as cooling devices are seeing structural demand growth. UBS noted that the European heat wave has already pushed spot power prices higher as aging electrical infrastructure struggled to cope with surging cooling demand — identifying ABB, Schneider Electric, and Siemens as key beneficiaries of the grid modernization investment that follows.

The investment thesis is converging on the same signal the weather forecast has been delivering for weeks: extreme heat is no longer a tail risk. It is a structural condition that requires structural solutions.

ERaaS Health is one of those solutions — at the intersection of AI, health insurance, and climate resilience. The market is beginning to understand what that’s worth.

What the Heat Means for Consumer Demand

For retailers, the Fourth of July heat signal is the demand catalyst that the first three weeks of June didn’t deliver.

Cooling categories are the immediate beneficiary — air conditioners, fans, portable cooling units. The current signal is strong enough to drive meaningful category lift across the Eastern US through July 13th.

Home center retail is where the weather impact is most apparent in July. The same heat that suppresses outdoor project activity in the short term drives purchases of cooling equipment, irrigation, and lawn recovery as consumers respond to the stress on their homes, yards, and vehicles.

One important caveat: electricity prices are elevated — driven by the Iran conflict and structural demand from AI data centers competing for the same grid capacity.

A heat wave that drives around-the-clock air conditioning use is not just a demand driver for cooling products. It is a direct wallet squeeze on an already financially stressed consumer. Every dollar going to the electricity bill is a dollar not going to discretionary retail.

Heat plus high electricity costs is a consumer squeeze, not a clean demand driver.

Why the Heat Moderates in the Second Half of July

The forecasters at NOAA’s Climate Prediction Center explain the science in their weekly technical discussion.

Here’s the plain-language version —

Two large-scale weather patterns are competing for control of the US atmosphere in mid-July. A ridge of high pressure — a dome of warm sinking air — is building strongly over the Western US. That is what keeps the West hot.

To the east, a trough — a dip in the upper-level wind pattern that allows cooler air to push southward — is expected to develop near the Great Lakes. That trough moderates heat in the Central and Eastern US during the second half of July.

The western ridge is clear across all forecast models — high confidence. The eastern trough is where the uncertainty lives. The models disagree on exactly where it sets up and how deep it goes. When models disagree at two to three weeks’ range, NOAA’s forecasters hedge — which is why the Central Plains and Great Lakes show Equal Chances in the Week 3-4 outlook rather than a strong temperature signal.

The same trough that moderates the heat also generates precipitation. Troughs are associated with rising air, clouds, and storm systems. The above-normal precipitation signal returning to the Northeast and Mid-Atlantic in the second half of July is a direct consequence of the troughing pattern expected to develop over the Great Lakes.

The bottom line: the Fourth of July heat event is real and building. The atmospheric setup driving it is not forecast to persist through mid-July. One strong demand week does not replicate last summer’s sustained heat pattern. The second half of July looks more like a normal summer — and normal after last July’s historic heat is a headwind, not a tailwind.

The G2 Weather Intelligence Signal

The Fourth of July window is delivering what the forecast promised weeks ago. Cooling categories and home center retail stand to benefit from a strong heat signal across the Eastern US through July 13th.

Watch the precipitation outlook as the holiday weekend comes into the short-range forecast. Rain is the swing variable — and the Week 3-4 maps show it returning to the Northeast and Mid-Atlantic in the back half of the month.

One more thing: I’ll report back on the hand fans and cooling fans. Wish us luck.

© 2026 G2 Weather Intelligence. All rights reserved.

One analytics foundation powers it all.

The weather-to-demand relationships established in seasonal planning — the historical correlations between temperature, precipitation, geography, and consumer behavior — are the same relationships that inform demand forecasting, inventory distribution, markdown pricing, and marketing activation. Build the analytics layer once. Every function draws from it.

The sum is far greater than the parts.

A retailer running weather-optimized marketing on top of weather-optimized inventory is compounding the signal. The demand is there because the forecast predicted it. The inventory is there because the distribution responded to it. The consumer shows up because the marketing was activated at exactly the right moment. That is the integrated system. That is the competitive advantage.

Fort Campbell, Kentucky. Early 1990s.

We had just returned from Operation Desert Storm — one of the most decisive military victories in American history. One hundred hours of ground combat. Coalition forces overwhelmed a battle-hardened Iraqi army with precision and speed the world had never seen.

I was the weather operations chief at Fort Campbell, Kentucky — deployed alongside the 101st Airborne Division, the 5th Special Forces Group, and the 160th Special Operations Aviation Regiment. The Night Stalkers.

If you know the US Army, you know what that means. These are not ordinary units. They’re among the most elite combat units in the American military — the ones that go first, go deep, and go where no one else can. The tip of the spear. Literally.

Our customers didn’t ask about the weather out of curiosity. They asked because lives depended on the answer. Go or no-go. Route selection. Insertion timing. The difference between a successful mission and a catastrophic one could be a single weather decision made hours in advance with incomplete information. We provided the intelligence. The warfighters made the call.

Back at Fort Campbell, in those days — before the internet, before digital everything — the weather station was still powered by teletype machines for data and fax machines for weather maps. It was all paper.

The only full-color satellite and radar loops we had came from The Weather Channel, on the TV mounted above the forecast counter. Functional between commercial breaks and, of course, Weather on the 8s.

The same Weather Channel that, twenty years later, would hire me to help transform it from a media company into a data and analytics platform.

I never imagined that one day I’d work there. But the question that Desert Storm planted never left me: What is weather intelligence actually worth if it doesn’t improve operational outcomes?

In 2011, The Weather Channel Company hired me to find out.

Paul Walsh, a former meteorologist with the U.S. Air Force who spent a little more than a decade in the private sector helping companies predict how weather patterns might affect their businesses, was tapped as the cable network’s vp of weather analytics. In the newly created role, Walsh will head up a nascent initiative to sell highly targeted weather predictions to the channel’s advertisers.

Source: ADWEEK

The Mission

My mandate in 2011 was to lead the development of a Weather Analytics function — and, with it, the transformation of The Weather Channel Company from a media company with data to a data-and-analytics company that was also a media company.

That distinction matters. A media company with data uses weather to attract an audience. A data and analytics company uses weather to change how businesses make decisions.

The forecast wasn’t the product. What you could do with it was the product.

Tampa, 2012

Not quite a year into the role, we ran our first live test of what would eventually become a new ad targeting system called WEATHERfx.

The Republican National Convention was about to begin in Tampa, and inside the newsroom, there was growing concern that a tropical system (nascent Hurricane Issac) developing in the Gulf could disrupt the event.

Producers were preparing contingency plans. Talent was standing by. If confidence in Tampa's potential impact increased, they would send an on-air team into the city days before landfall.

One of our first advertising partners was Duracell. Instead of waiting for the National Weather Service to issue official warnings, we asked a different question: what if we used the forecast — not the weather itself — to decide when to begin advertising?

As confidence increased that Tampa could be affected, we triggered the Duracell campaign days before official warnings were issued. Consumers hadn’t started preparing yet — but they would. Batteries would be one of the first things they bought when they did.

The campaign dramatically outperformed expectations and, as I recall, led to a significantly larger Duracell commitment going forward.

But the bigger lesson wasn’t about batteries. We hadn’t just run a better ad. We’d answered a question that most marketing organizations never ask: when is the consumer most likely to need this product — and how far in advance can we see it coming?

Weather wasn’t just something to forecast. It was a marketing signal.

The Four Questions Every Company Is Trying to Answer

Every retailer, brand, and consumer-facing business is trying to answer the same four questions — usually independently, rarely simultaneously.

How much do I need? Seasonal planning. The weather predicts demand deviation from baseline before the season begins. The seasonal buy is a fixed decision — but it should be a probability-weighted one, informed by the best available long-range forecast at the time of the buy. Get this wrong and every downstream decision is fighting an uphill battle.

Where should I put it? Distribution. The right inventory in the right stores at the right time. A heat signal building in the Northeast is a reallocation trigger for cooling categories — before the heat arrives, not after.

What should I charge for it? Markdown pricing. The same weather signal that drove the inventory decision should inform the pricing decision.

• A cool April with a warm pattern building in the 10-day forecast is a hold-the-markdown signal for spring apparel — the demand is coming, don't surrender the margin early.

• A cool April with no recovery in the forecast is a mark-it-down signal — move the inventory before the season closes. The consumer's price sensitivity shifts with the weather. The pricing model has to see the same forecast as the inventory model.

When should I tell customers — and what should I say? Marketing activation and creative are two sides of the same decision, and the weather signal drives both.

Timing first. Don’t advertise umbrellas after it starts raining. Activate when forecast confidence crosses the threshold that drives consumer preparation behavior. The Duracell campaign worked because we acted on the forecast — not the event. By the time the NWS issued warnings, every competitor was running the same message. We were already three days ahead.

Message second. Cold weather, heat wave, storm prep, backyard weekend, holiday travel — different context, different mindset, different message.

A consumer in Boston during a heat wave and a consumer in Phoenix during the same event have completely different baseline experiences. The Boston consumer is caught off guard. The Phoenix consumer is in their normal summer.

Same temperature signal. Completely different creative brief. The forecast tells you when to activate. The weather regime tells you what to say. Both inputs come from the same signal.

The Big Idea

None of these decisions should happen independently.

A markdown system that doesn’t see the same forecast as the marketing team is working against itself. A marketing campaign that activates on a calendar date while the inventory system is responding to a weather signal is firing blind. A demand forecast that updates weekly while the promotional calendar updates monthly is leaving money on the table.

The weather signal connects all four. Inventory. Distribution. Pricing. Marketing. One signal. One system. Every function responds to the same forecast at the same time.

That signal today is the weather. Tomorrow, it’ll be AI-generated demand intelligence that incorporates weather and a dozen other real-time variables. The principle remains exactly the same: the value isn’t in having better data. It’s in getting every function to act on the same data at the right time.

The Research Confirms It

Two studies make the case — one from academia, one from the organization I helped build.

A peer-reviewed study published in the INFORMS journal Marketing Science — drawing on field experiment data from over six million mobile users across 344 cities — found that consumer response to mobile promotions was 1.2 times higher and 73% faster in sunny weather than in cloudy weather.

More importantly, people responded especially well when actual weather was better than forecast — reinforcing that forecast context matters more than current conditions. You are not marketing to the weather. You are marketing to the consumer’s expectation of the weather.

In 2025, The Weather Company took that finding further. Working with Neuro-Insight, a global leader in brain-based marketing research, they conducted a neuroscience study — “Wired for Weather” — that measured subconscious brain responses in 182 participants exposed to weather conditions and paired advertisements. Key findings:

Overall:

• Ads aligned with weather-driven emotional states lifted ROI by up to 18%

• Ads served immediately after a consumer checks the weather tap into fresh emotional responses — a more powerful moment of influence than calendar-based or demographic-targeted campaigns

Sunny days:

• +10% Engagement

• +12% Detail Memory — a key predictor of purchase intent

• +19% Global Memory — brand storytelling retention

Rainy days:

• +29% Engagement

• +22% Detail Memory

• +25% Global Memory

Gen Z:

• Up to +13% higher memory encoding

• +12% increased engagement on sunny days compared to the general population

The Weather Company built the capability I helped start into a precision advertising platform backed by neuroscience. The Duracell campaign in Tampa was the proof of concept. The “Wired for Weather” study is the scientific validation.

Weather isn’t just a backdrop. It is an emotional and cognitive signal that shapes how people process and remember brand messages. The forecast changes the consumer’s mindset. The right message in the right weather regime doesn’t just reach the consumer — it reaches them when they are most neurologically receptive to it.

Where This Fits

Marketing is the fourth post in this series for a reason. It is not the first lever you pull. It is the demand activation layer — the mechanism that ensures that when inventory is in the right place and priced correctly, the consumer actually shows up to buy it.

Seasonal planning sets the buy. Demand forecasting optimizes the distribution. Pricing manages the margin. Marketing activates the demand. The four are not four separate decisions. They are one decision made at four levels of the planning hierarchy — and the same weather signal connects all of them.

A retailer running weather-optimized marketing on top of weather-optimized inventory is compounding the signal. That compounding effect is where the real competitive advantage lives. Not in any single capability — in the integration of all of them.

The Forecast Was Never the Product

In 2015, IBM announced the acquisition of The Weather Company for a reported $2 billion.

Not for the television network. Not for the consumer app. For the data, the analytics, and the enterprise intelligence platform built on the thesis that weather was a business signal — and that the companies that could act on it faster and more precisely than their competitors had a structural advantage.

At Fort Campbell, we called that advantage decision superiority. The force with the best weather intelligence didn’t just have a better forecast — it had better timing, better positioning, and better coordination across every function of the mission.

The forecast was the input. Synchronized action was the output.

WEATHERfx applied the same principle to marketing. The Duracell campaign wasn’t remarkable because we had better weather data than our competitors. It was remarkable because we connected the forecast to the activation decision faster than anyone else — and marketing responded to the signal before consumers even knew they needed the product.

IBM paid $2 billion in part for that vision. The full realization of it — every function of a retail or consumer enterprise responding to the same weather signal simultaneously — remains, frankly, a work in progress across the industry.

That gap is precisely what this series is about.

Next in the series: weather-driven staffing optimization — scheduling for the forecast, not the calendar.

Watch the Conversation

How Weather Influences Retail, Marketing, and Human Behavior — I joined David G. Ewing on Content Kingdom to talk through the ideas behind this post: weather intelligence, consumer behavior, predictive marketing, and where AI takes it next. From strawberry Pop-Tarts before hurricanes to the future of weather-driven advertising.

© 2026 G2 Weather Intelligence. All rights reserved.

This is a heat event arriving just in time for the Fourth of July holiday weekend — lining up almost exactly with the anniversary of last June’s historic heat dome.

This is how the forecast system works. Longer-range outlooks establish the probabilistic envelope. Short-range forecasts fill in the details. The signal wasn’t absent before — it simply hadn’t resolved. Now it has.

The Anniversary Signal

Last June ranked 123rd warmest out of 131 years in the Northeast. The heat dome pushed Boston to 102 degrees, Philadelphia to 101, Newark to 103. Nearly 160 million Americans were under heat alerts simultaneously.

The current forecast is not last June. But the above-normal signal now building across the Northeast and Mid-Atlantic through the Fourth of July window — arriving as the Northeast returns to near-normal after a volatile spring and early summer — has the makings of a demand catalyst for cooling categories, outdoor recreation, and shore traffic.

The swing variable, as always, is the rain. Watch the precipitation outlook as the holiday weekend comes into the short-range forecast range.

What Last Summer Cost New York City

Last summer wasn’t just a retail story. It was a sustained public health emergency that ran from late June through mid-July.

The first event — June 22 through June 25 — was the most extreme. Central Park reached 99 degrees, breaking a record that had stood since 1888. JFK Airport hit 100 degrees for the first time since 2013. Newark hit 103. More than 150 million people across the Eastern US woke up to heat warnings.

Emergency departments in New York City reported 341 heat-related visits during the four-day event, with 141 on a single day — the highest one-day total since 2017. Nineteen people died. All but three were Black or Latino. Most died at home, without air conditioning.

That was just the beginning.

NYC experienced four separate heat waves over the course of the summer. A fourth heat wave arrived the week of July 11, with a Heat Advisory in effect July 16-17 as heat index values reached 100-104 degrees. By the end of the season, NYC finished 2025 as its fourth-warmest year on record and its second-hottest summer on record.

The NYC Health Department's 2026 annual report found that approximately 500 New Yorkers die from heat-related causes each year — most from chronic conditions worsened by heat. Of those, deaths from direct heat stress are far rarer.

Last summer's June heat wave alone caused 19 of them — well above the annual average of five — and disproportionately affected Black and Latino New Yorkers.

As the Fourth of July heat signal builds across the Eastern US this week, both lenses matter.

AI-First Health Resilience — At the Speed of a Heat Wave

Those deaths are preventable. Last summer’s heat events showed what prevention looks like when technology moves at the speed of the forecast.

ERaaS Health, founded by Samrat Kulkarni, is developing AI-first health intervention tools specifically targeting heat-related illness and mortality risk. I serve as an advisor to the company. During last summer’s heat events, ERaaS demonstrated what proactive, technology-enabled intervention can look like in practice.

Through scalable outreach, the system moved from data to proactive intervention in under 24 hours — identifying at-risk members, initiating rapid engagement via an AI agent, and seamlessly surfacing escalations to health plan care teams for targeted support, including wellness checks, cooling center referrals, and emergency interventions.

Every interaction was captured in real time, creating a closed-loop system connecting escalation detection, rapid outreach, and outcome tracking.

The results among engaged members were striking: a 47% reduction in ER and inpatient admissions, 9 out of 10 members’ call satisfaction, and an estimated 5-to-1 return on investment.

As heat events become more frequent and more severe — in the US and globally — the need for this kind of proactive, technology-enabled health resilience infrastructure moves from an interesting idea to a critical public health imperative. The ERaaS model points toward what that infrastructure looks like at scale.

What July Is Showing

The NOAA Official 30-Day Forecast, issued on June 18, covers the full month of July. It tells a tale of two coasts — and significant uncertainty in between.

The West carries a strong above-normal temperature signal — likely above normal across the Pacific Northwest, California, and the Rockies, with 70-80% probability in the core.

For Seattle and Portland, already structurally underprepared for sustained heat, this is the signal that matters most for health risk and cooling demand.

The East is more nuanced. The Mid-Atlantic and Southeast show a modest warm lean. New England and the Great Lakes sit in Equal Chances for the month as a whole. That is not a strong demand driver against last summer’s sustained heat pattern — but it is not a cold signal either.

The precipitation map adds the swing variable.

Above normal precipitation is forecast across the Ohio Valley and into the Central Appalachians through July. Warm and wet conditions suppress outdoor demand even when the temperature signal is positive.

The coastal Northeast — where shore traffic and outdoor recreation demand concentrates during the Fourth of July window — sits in Equal Chances, which is the better news in this map for retailers with exposure to that geography.

July outlooks carry more uncertainty than short-range forecasts — this is normal for summer at the monthly range when large-scale atmospheric patterns are less persistent than in winter. The directional signal is consistent with what the short-range maps are already showing: above normal in the West, modest and uncertain in the East.

One event around the Fourth of July does not replicate the sustained demand effect seen last summer. How the pattern evolves over the rest of July will signal whether retail results hold up against last summer’s historic comparison.

The Consumer Headwind That Won’t Move

To materially move consumer behavior in summer, you need roughly two standard deviations from normal sustained over weeks. Last June, cleared that bar by a wide margin. The current forecast is above normal — but not yet at the outlier threshold that generates the demand surge retailers need to lap last year’s results.

And the consumer absorbing this heat is not the same consumer who carried last summer. Electricity prices are elevated — driven by the Iran conflict, which is pushing energy costs higher, and by structural demand from AI data centers competing for the same grid capacity.

A heat wave that drives around-the-clock air conditioning use is not just a demand driver for cooling products. It is a direct wallet squeeze on an already financially stressed consumer. Every dollar going to the electricity bill is a dollar not going to discretionary retail.

Heat plus high electricity costs is a consumer squeeze, not a clean demand driver.

Europe Is Already There

While the US watches a heat signal build for the Fourth of July window, Western Europe is already in the midst of it.

Temperatures in parts of France exceeded 104 degrees Fahrenheit this week, drawing comparisons to the catastrophic 2003 heat wave that killed 70,000 people across the continent. Britain’s Met Office issued extreme heat warnings. France put much of the country under red-level alert.

The New York Times reported this morning that of France’s 52 official heat waves since 1947, half have occurred in the past 16 years — an acceleration that epidemiologists are directly linking to climate change. “We have adapted, but it is far from enough for what is coming,” said Pierre Masselot of the London School of Hygiene & Tropical Medicine.

Europe built early warning systems and cooling shelters after 2003. They helped. But air conditioning penetration remains low — roughly 25% of French homes have cooling units. And the absolute number of senior citizens across the EU has risen by 40% over the past two decades, expanding the most vulnerable population at the same time that heat waves are accelerating.

This is the crash test, as one French epidemiologist put it. The preparation has helped. It has not solved the problem.

The Fingerprints of Climate Change

The European heat wave and the building US heat signal are not unrelated weather events. They are fingerprints of a warming climate system in which extreme heat events are becoming more frequent, more intense, and more economically and medically consequential.

El Niño adds a natural amplifier atop an already warming baseline — raising the ceiling on what extreme heat events can deliver this summer and into the winter season.

The 2021 Pacific Northwest heat dome killed hundreds in a region structurally unprepared for sustained heat. The 2003 European heat wave killed 70,000. Last June’s US heat dome put 160 million Americans under heat alerts. This week, Europe is in the middle of its next major test.

The G2 Weather Intelligence Signal

For retailers: the Fourth of July heat signal is real and building. Cooling categories, outdoor recreation, and shore traffic may get a demand boost. But watch the precipitation outlook — rain is the swing variable — and don’t mistake a one-week heat event for the sustained two-standard-deviation anomaly that drives the kind of July surge the retail home sector saw last year.

The weather backstop that carried spring retail sales performance is still absent at the macro level. This heat event is a local and temporal signal, not a national tailwind.

Musical Coda

© 2026 G2 Weather Intelligence. All rights reserved.

El Niño was officially declared on June 11 — and a Super El Niño may be on the way.

NOAA’s Climate Prediction Center upgraded its alert status from El Niño Watch to El Niño Advisory, signifying that El Niño conditions are now present in the equatorial Pacific and expected to strengthen.

My pal and G2 Weather Intelligence subscriber Andrew Freedman covered it for CNN — if you follow weather and climate, he’s a great read.

NOAA now puts the odds of a strong El Niño at nearly 90% and the odds of a very strong — Super — El Niño at over 60% by autumn. Some models project that Pacific sea surface temperatures will exceed the record-breaking levels of 1982-83 and 1997-98.

If verified, this would be the most powerful El Niño in recorded history.

One important caveat. The oceanic signal and the atmospheric response are not the same thing. NOAA’s own forecasters noted this week that El Niño’s teleconnections to North American weather are weak during summer — meaning the near-term forecast is being driven by model guidance, not El Niño composites.

The full force of El Niño arrives in winter. I’ll be covering the balance of summer implications in a separate post. This one is about what history tells us about the holiday season — and what retailers should be doing about it right now.

What History Says — The 2015-16 Playbook

The most recent comparable El Niño was 2015-16 — strong, though not at the scale now being forecast for this event. The retail record from that winter is instructive.

Outerwear sales in the US fell 10% during the 2015 holiday season. Shoppers walked mall parking lots in shorts and short sleeves with no reason to browse winter apparel. Retail consultancy Planalytics estimated apparel retailers lost approximately $838 million in sales from mild November and December weather alone — and that figure didn’t include department stores or discount retailers.

Macy’s CEO Terry Lundgren put a precise number on it in the company’s official holiday sales release: “About 80 percent of our company’s year-over-year declines in comparable sales can be attributed to shortfalls in cold-weather goods such as coats, sweaters, boots, hats, gloves and scarves.”

Comparable sales on an owned basis fell 5.2% for the November-December period. The culprit was historically warm weather in the northern climate zones where Macy’s and Bloomingdale’s are most concentrated.

By late December, with markdowns above 50% on winter apparel, industry researchers noted it would be very difficult to make up margins in the coming months. A last-minute cold snap in the final two weeks of December generated an estimated $200 million in apparel sales — described by one analyst as “better than a sharp stick in the eye” but nowhere near enough to save what he called “an otherwise sorry season.”

The 2015-16 El Niño was strong but not historically strong. The event now building is forecast to potentially exceed it. The retail implications scale accordingly.

The 1997-98 Comparison

Go back to the last truly historic El Niño — 1997-98 — and the picture is more nuanced but equally instructive.

A peer-reviewed study published in the Bulletin of the American Meteorological Society found that the 1997-98 El Niño produced an estimated $4 billion in direct losses nationally but approximately $19 billion in economic benefits — a net positive for the US economy overall.

The northern half of the country experienced warmer winters, record retail sales of seasonal products and homes, record construction activity, and significant savings on heating costs.

The net number was positive, but the distribution was anything but uniform. Heating fuel costs dropped significantly. Home improvement and construction benefited from the mild winter. Outerwear, winter recreation, and snow-removal equipment incurred losses.

The Category Winners and Losers

The pattern across both events is consistent. During the 2015-16 El Niño winter, residential heating oil prices fell approximately 29% compared to the prior year, and households paid 15% less for propane — freeing up consumer wallet share that might otherwise have gone to heating costs.

That savings offset is real and worth noting for retailers in non-weather-sensitive categories.

But for anyone selling outerwear, heating equipment, ski gear, or cold-weather seasonal products, the El Niño winter is a direct headwind against demand that no amount of promotional activity can fully overcome.

What a Potentially Historic El Niño Means for Holiday 2026

If the current forecast verifies — strong to very strong (super?) El Niño by autumn, with the full atmospheric coupling arriving for the November through January selling window — the historical pattern points to several clear implications:

• Warmer-than-normal temperatures across the northern US and Great Lakes reduce demand for outerwear, heating equipment, cold-weather accessories, and winter recreation products. These are not marginal effects. In 2015-16, a merely strong El Niño cost specialty apparel retailers nearly $1 billion in lost sales.

• The South and Southeast face wetter than normal conditions — suppressing outdoor activity and driving some demand toward indoor entertainment and home categories.

• Home improvement and construction benefit from the mild winter — a tailwind for Home Depot and Lowe’s comparable to what the warm spring delivered in Q1 2026.

• Heating fuel cost savings free up consumer wallet share — a modest positive for general merchandise and discretionary categories, particularly among lower-income consumers for whom energy costs represent a larger share of household spending.

The overall picture: a warmer El Niño winter is not a retail disaster. It is a category rotation.

The retailers who plan for it in advance — shifting inventory away from cold-weather categories and toward the categories that benefit — will outperform. The ones who carry their historical cold-weather buys into a warm El Niño winter will be the ones citing weather on their Q4 earnings calls. After the fact. Without a plan.

This Is Exactly What an Integrated Weather Strategy Is Built For

The 2015-16 El Niño was forecast with high confidence months in advance. NOAA called it in the summer of 2015. The retailers who acted on that forecast — adjusting their holiday buys, managing outerwear inventory more conservatively, and leaning into home improvement and indoor categories — had a materially better holiday season than those who didn’t.

The retailers who ignored it were the ones blaming the weather on their January earnings calls.

The G2 Weather Intelligence Brainstorm series is built around exactly this kind of end-to-end weather strategy — from the seasonal buy to the markdown decision to the marketing activation. Three posts are live. Five more are coming. The capstone shows what it looks like when every piece is running simultaneously.

The forecast is calling for a potentially historic (Super?) El Niño winter. The retailers who start building their weather strategy now will be better positioned than the ones who read about it in a Q4 earnings call transcript next February.

Video Coda

© 2026 G2 Weather Intelligence. All rights reserved.

This is the third post in the G2 Weather Intelligence Brainstorm series on building an integrated retail weather strategy.

The first post covered seasonal planning — how much to buy. The second covered demand forecasting and distribution — where to send it. This post covers pricing — specifically, markdown pricing — and the weather variable that most optimization systems still don’t fully exploit.

A Story From the Field

Years ago, I worked directly with a senior executive at one of the country’s largest women’s specialty apparel retailers. The company had invested in a sophisticated price optimization system — it analyzed sell-through rates, weeks of supply, calendar position, and historical markdown performance.

State of the art in every respect except one: it had no way to see the weather coming.

That gap is where I came in. I provided her with ad hoc weather intelligence the old-fashioned way, directly from my brain, and she used it to override the system’s recommendations whenever the forecast told a different story than the sell-through data.

The results were striking. One particular spring stands out. It started cooler than normal; inventory turns slowed, and the optimization system reviewed the sell-through data and recommended a 40% markdown to clear seasonal product.

The weather forecast told a different story. Temperatures were about to shift. A warmer pattern was building. She hedged — took a 20% markdown instead of the system’s recommended 40%.

When the weather turned and demand returned, the inventory cleared. At 20% more margin than the system would have surrendered.

That is not a rounding error. On a seasonal buy of any size, a 20-point difference in markdown depth is real money. And it was being left on the table — every season — because the pricing model was optimizing against incomplete information.

Weather Changes What Consumers Will Pay

The umbrella example is the simplest illustration. A rainy day in New York City sharply increases demand for umbrellas. Supply is fixed. The consumer who needs an umbrella right now is not price sensitive — demand is inelastic.

The same umbrella on a sunny day in the same store is a discretionary purchase. The consumer can wait, shop around, or simply not buy. Demand is elastic. Price sensitivity is high.

Same product. Same store. Different weather. Completely different pricing power.

The apparel example works in reverse. A cool April makes spring dresses optional — consumers don’t need them yet. She can wait. Demand is elastic, and price sensitivity is high. A warm April makes the same dress feel urgent. Pent-up seasonal demand releases. The consumer converts at higher prices.

Academic research confirms the relationship directly. During periods of extreme heat, consumers have low price elasticity for cooling products due to rigid demand — price sensitivity drops and margin opportunity rises. When temperatures moderate, the same products become highly elastic — consumers will wait, shop around, or simply not buy.

Weather doesn’t just change how much consumers buy. It changes how much they will pay — and how sensitive they are to price changes. That is the definition of price elasticity. And weather is one of the most powerful real-time drivers of elasticity that most markdown optimization systems don’t fully account for.

Coca-Cola Understood This in 1999

In 1999, Coca-Cola quietly began testing a vending machine that could automatically raise prices for its drinks in hot weather. CEO M. Douglas Ivester described the logic plainly: “Desire for a cold drink increases during a sports championship final held in the summer heat. So it is fair that it should be more expensive.”

The concept was economically sound. The machine was withdrawn after significant customer backlash — consumers understood the logic and rejected it anyway. But the underlying principle has never been wrong.

The Harvard Business School published a case study on the episode in 2000. It’s been taught in business schools for 25 years. Weather-driven price elasticity is real, measurable, and exploitable. The question is how to apply it in a way that optimizes margin without triggering the Coca-Cola backlash.

The answer for retail is not raising prices in a heat wave. It is avoiding unnecessary markdowns when a weather-driven demand recovery is forecast. That is the women’s apparel executive’s insight applied at scale.

Where the Industry Actually Is

The leading retailers are moving in the right direction. Walmart has begun incorporating weather forecasts into pricing decisions — adjusting seasonal product prices earlier than historical calendar-based approaches would have dictated, based on regional weather outlook data.

The academic case for why this matters is striking.

A 2024 peer-reviewed study looked at how much of the week-to-week swing in retail sales a standard forecasting model — one built on past sales, day of week, and promotions — simply couldn’t explain. When weather data was added to the same model, it accounted for up to an additional 47% of the unexplained variation for individual products and up to 56% for entire product categories.

In plain terms, weather isn’t a minor adjustment to a forecast. For some products, it explains nearly half of the swings in demand that a non-weather model has no way to see coming. That is the difference between a forecast that is roughly right and one precise enough to drive real inventory and pricing decisions.

But implementation is uneven. And the gap between claiming to use weather and using it precisely is significant.

Consider this: a major home improvement retailer’s CFO cited “cold, wet weather in May” as the reason for weaker sales in a recent quarter. Independent weather data showed May was wet but not cold. The retailer was using weather as a narrative — not as a signal.

That is the real problem. Weather intelligence is not a single capability. It is a connected system — seasonal planning, demand forecasting, distribution, pricing, marketing — each layer informed by the same forecast signal, each layer building on the one before it.

A retailer who uses weather for inventory planning but not pricing is leaving margin on the table. A retailer who uses weather for pricing but not marketing is missing the demand activation window. A retailer who uses weather for all of it — systematically, end-to-end — is building something closer to a weather-proof earnings model.

Why Most Markdown Systems Still Get This Wrong

Current markdown optimization systems have improved significantly. Leading platforms now list weather as a factor in their models. But listing weather as a variable and systematically integrating the forward forecast into real-time markdown decisions are different things.

The gap is not awareness. It is precision and integration. A system that incorporates historical weather patterns is not the same as one that compares the current inventory position to the 10-day forecast and adjusts the markdown recommendation in real time as the forecast evolves.

The women’s apparel executive was doing the latter — manually, in her head, one season at a time.

The Weather-Informed Markdown Framework

Three inputs the markdown model needs to use the weather signal effectively:

Current weather anomaly. What has the weather done to demand velocity in recent weeks? Is inventory turning slower because of a weather headwind or because of a product problem? These are different diagnoses that require different responses. A weather-driven slowdown is temporary. A product problem is not.

Forward forecast. What does the next 6-30 days look like for the categories with excess inventory? Is the demand headwind temporary or persistent? A cool week followed by a warm pattern is a hold-the-markdown signal. A cool month with no recovery in the forecast is a mark-it-down signal.

Elasticity by weather regime. How price-sensitive are consumers in this category under current and forecast conditions? A heat wave makes cooling products inelastic — hold price, maximize margin. A cool June makes them elastic — mark down earlier and deeper to move product before the season closes. The markdown depth should reflect the elasticity regime, not just weeks of supply.

The Digital Price Tag Enabler

Electronic shelf labels and digital price tags make weather-responsive pricing operationally feasible at scale for the first time. Price changes that once required manual intervention across thousands of SKUs and hundreds of stores can now be executed automatically — in minutes, not days.

The technology infrastructure for weather-driven dynamic pricing exists today. The missing ingredient is connecting the weather forecast to the pricing engine in real time.

Agentic AI as the Execution Layer

The markdown decision requires holding three variables simultaneously — current inventory position, forward weather forecast, and real-time elasticity estimate — and updating the response as each variable changes.

That is exactly the multi-variable, continuously updated optimization problem that agentic AI was built to solve.

The agent monitors the forecast. When the signal changes — a warming pattern arriving sooner than expected, a cold snap extending further than forecast — it adjusts markdown recommendations before the demand window opens or closes. No manual override required.

What the women’s apparel executive did in her head, one season at a time, the agent does automatically — across every category, every store, every week — at the speed the forecast window requires.

From Override to Automation

The women’s apparel executive was doing manually what the system should have been doing automatically. She had the weather intelligence. She had the inventory position. She made the call — and recovered 20 margin points the system would have surrendered.

The difference between her approach and the fully integrated framework is not the insight. It is scale and speed. One executive can override one system for one season. An agentic AI holds the weather signal, the inventory position, and the elasticity estimate simultaneously — and executes before the opportunity closes.

The markdown model without weather is optimizing against incomplete information. That is not a pricing problem.

The tools exist. The data is publicly available. The gap is the absence of an end-to-end weather strategy — one that connects seasonal planning, demand forecasting, distribution, pricing, and marketing into a single integrated system, each layer informed by the same forecast signal, each decision building on the one before it.

That is what separates a retailer that occasionally benefits from favorable weather from one that has built weather intelligence into the foundation of its operations — and earnings.

Next in the series: weather-driven marketing activation — timing campaigns to the forecast rather than the calendar.

© 2026 G2 Weather Intelligence. All rights reserved.

This is the second post in the G2 Weather Intelligence Brainstorm series on using weather intelligence across the retail planning cycle.

The first post covered seasonal planning — using long-range weather outlooks to set the strategic foundation before the season begins.

This post covers the next step: demand forecasting and inventory optimization. If the first post was about strategy, this one is about money.

Let’s start with a thought experiment.

The academic and industry research is unambiguous. Incorporating weather data into demand forecasting models reduces forecasting error by up to 45% for seasonal product categories.

AI-driven demand forecasting delivers average inventory reductions of 20-30%, generating working capital improvements of $15-20 million per $1 billion in revenue annually.

Better weather-aligned inventory positioning delivers a 20-70 basis point profit lift through higher full-price sell-through, lower inventory costs, and optimized markdown timing.

And the baseline math is unforgiving — inventory carrying costs run 20-30% of inventory value per year, meaning excess seasonal inventory is not just a markdown problem. It is a capital problem.

Now apply those numbers to three retailers I’ve been covering all earnings season:

• Home Depot generates approximately $160 billion in annual revenue. Roughly 40-50% is weather-sensitive — lawn and garden, outdoor living, cooling, heating, building materials. At $15-20 million in annual working capital improvement per billion in weather-sensitive revenue, the potential annual benefit for Home Depot alone is nearly $1 billion to $1.6 billion.

• Lowe’s generates approximately $83 billion in annual revenue with a similar category mix — roughly $33-42 billion in weather-sensitive revenue. Annual working capital improvement potential: $495 million to $840 million.

• Walmart’s seasonal category exposure — apparel, lawn and garden, outdoor living, holiday, automotive — conservatively represents $100 billion in weather-sensitive revenue. Annual working capital improvement potential: $1.5 billion to $2 billion.

Combined, these three retailers represent a potential $3 to $4.4 billion in annual benefit from a publicly available data source accessible to any competent data science team today.

This is a thought experiment, not a guarantee. The research establishes the range. The actual outcome depends on execution, category mix, and the degree to which the weather signal is integrated into planning systems.

But the order of magnitude? Not in dispute

Obvious in Theory. Hard in Practice.

If it were easy, everyone would be doing it. The reason most retailers are not capturing this value is not a lack of awareness. It is the genuine complexity of the weather-to-demand relationship.

Weather doesn’t drive demand in a simple, linear way. To use the signal effectively, you have to resolve several problems that are harder than they sound:

What actually happened? Before you can forecast weather-driven demand deviation, you need to accurately measure what the weather has historically driven. That means isolating the weather component from all other variables — promotions, price changes, competitive activity, calendar shifts, consumer sentiment — at the category and SKU levels.

Lag effects. This is one of the biggest complexities in weather-driven demand modeling. Weather doesn’t always drive demand in the same week it occurs — and the lag structure varies dramatically by category.

A cold snap in April immediately suppresses spring apparel sales. But that demand doesn’t disappear — it defers. When temperatures normalize, pent-up demand is released, often in a concentrated burst that can be mistaken for an execution win rather than a weather catch-up.

A sustained heat wave drives air conditioner and fan sales in week one. That same heat places stress on vehicle batteries, belts, and cooling systems — and that stress shows up as automotive parts and repair demand two to three weeks later, after the damage has accumulated.

Get the lag structure wrong, and the model is right about what will happen but wrong about when — triggering inventory responses too early or too late, leaving you understocked when demand peaks and overstocked when it doesn't.

Weather forecast-driven demand. Consumers don’t just respond to weather. They respond to weather forecasts. A heat wave widely forecast three days out drives air conditioner sales before the heat arrives. The model has to account for the demand the forecast generates, not just the demand the weather event itself drives. These are different signals.

Calendar context. A 5-degree above-normal temperature anomaly in early June drives a materially different demand response than the same anomaly in late July. July heat feels normal. June heat feels early — and early heat drives a disproportionate demand response because pent-up seasonal demand is still building. The model has to understand where it is in the seasonal calendar, not just what the temperature is doing.

Geography and microclimate. A national average temperature anomaly is nearly useless for store-level inventory decisions. A heat dome in the Northeast is irrelevant to a store in Phoenix, where it is always hot in June. The signal has to be specific to the store’s local climate, the consumer’s local baseline, and the category’s local demand profile.

This is why the problem has been discussed for decades and solved by very few.

The Data Is Already Available — Free

Here is what most retail analytics teams may not fully appreciate. NOAA’s Climate Prediction Center publishes probabilistic forecasts at no cost that extend a full year out, quarter by quarter.

These are scientifically grounded probability signals — not gut calls or rough estimates — updated monthly and precise enough to inform meaningful planning decisions at every horizon from near-term promotions to long-range seasonal buys.

Commodity traders, energy companies, and agricultural planners invest heavily in proprietary weather intelligence — specialized ensemble models and commercial vendor enhancements that provide incremental forecasting edges worth billions in trading profits.

Retailers don’t need that level of precision or investment. The NOAA probabilistic forecast data is more than sufficient for retail planning purposes — and it’s free.

The forecast hierarchy covers every planning horizon a retailer needs:

• 6-10 days: High confidence. Actionable for near-term promotional and staffing decisions.

• 8-14 days: Directional confidence. Actionable for inventory reallocation and reorder decisions.

• 30-day: Monthly outlook. Actionable for buy optimization and category-level positioning.

• Seasonal outlooks: Quarterly probability forecasts updated monthly, extending a full year forward. Actionable for long-range buy decisions and financial risk management.

Any data scientist worth their salt can access and integrate all of these data streams at no cost today. The monthly updates to the seasonal outlooks are particularly underutilized — each update is a signal to revisit buy quantities, adjust inventory positions, and reshape promotional calendars for the quarter ahead.

Most retailers treat the seasonal buy as a fixed decision. The probabilistic forecast treats it as a continuously updated signal. The gap between those two approaches is where the margin is lost.

There are also many commercial weather data vendors who provide this data in more accessible formats with additional value-added services. But the foundational data is free, publicly available, and ready to use.

Why Agentic AI Is the Unlock

Machine learning resolves the complexity of the weather-to-demand relationship. It can identify lag structures, calendar sensitivities, forecast-driven demand components, and geographic specificity simultaneously — and update those relationships continuously as new data arrives.

Machine learning alone produces a better model—agentic AI turns that model into operational action. The distinction is critical.

A model that tells a merchant “cooling category demand will be 15% above normal next week in the Northeast” is useful.

An agent that automatically triggers the reorder, adjusts the promotional plan, reallocates inventory from weaker-signal markets, and updates the markdown calendar — without waiting for a human to interpret the output and make a decision — is transformational.

The forecast window is too short and the store footprint too large for human decision-making to keep up. Agentic AI is the lever that finally makes this solvable at scale. The technology exists today. The forecast data is available and free. The constraint is building the system that connects them.

The Throttle — Inventory as the Constraint

The signal tells you what demand will do. The inventory position tells you what you can do about it. The system has to hold both simultaneously.

• A warm June signal with strong inventory depth in cooling categories is an opportunity — accelerate promotions, pull forward demand, maximize sell-through at full price.

• A warm June signal with thin cooling inventory is a different problem — reorder immediately, reallocate from weaker-signal markets to stronger ones.

• A cool June signal with heavy cooling inventory is the worst case — markdown acceleration, reorder suppression, promotional investment to move product before it ages.

AutoZone’s comp deceleration in the final two weeks of Q3 was not just a weather story. It was an inventory response story. The demand signal was visible in the temperature data two weeks before it showed up in the numbers. The system to act on it wasn’t in place.

The End State — Store-Level Intelligence at Scale

The ultimate expression of this framework is not a regional signal applied uniformly across a store fleet. It’s a store-level demand forecast — specific to the microclimate, the local inventory position, and the category mix of each individual location — updated continuously as the forecast evolves and as the monthly seasonal outlooks are revised.

Consider what that looks like in practice:

• A distribution center receives a reallocation signal because the 8-14 day outlook indicates a heat dome building in the Northeast.

• A store in Boston gets a promotional activation trigger for cooling categories three days before the heat arrives.

• A store in Chicago with excess spring apparel inventory receives a markdown signal because the 6-10 day outlook indicates below-normal temperatures through the week.

• The monthly seasonal update triggers a revision to the Q3 outdoor living buy before the purchase order is placed.

One weather data source. One model. Store-level precision at every planning horizon — from the distribution center to the shelf. Demand forecasting informs the decision. Inventory optimization shapes the response. Agentic AI executes before the opportunity closes or the markdown clock starts running.

From Signal to Action

The weather forecast signaled that AutoZone’s comps would soften two weeks before they did. The signal was visible. The system to act on it wasn’t in place.

Turning that signal into a store-level inventory and promotional response — automatically, at scale, accounting for lag effects, calendar context, forecast-driven demand, and local microclimate — is a hard problem. It has been discussed for decades. Agentic AI is what finally makes it solvable.

Demand forecasting and inventory optimization are one layer of a broader end-to-end weather strategy that spans seasonal planning, pricing, marketing, staffing, and financial risk management. Each layer builds on the one before it. Each one is powered by the same publicly available data source.

The question is not whether the signal exists. It is whether the system is built to act on it.

© 2026 G2 Weather Intelligence. All rights reserved.

"God created economists to make weather forecasters look good."

The most accurate forward-looking signal available to any retailer with seasonal categories isn't on Wall Street. It's at NOAA. And the latest outlook is unambiguous.

The updated 6-10 day outlook issued June 8 shows a deep below-normal temperature pool dominating the Central and Eastern US through June 14-18 — blue from the Rockies to the Mid-Atlantic. Above-normal precipitation across the South and Southeast.

The 8-14 day outlook shows the cold pool retreating but persisting across the Central US through June 16-22, with above-normal precipitation expanding across the East.

The pattern is clear. Cooler and wetter across the most densely populated and weather-sensitive retail markets in the country. The exact opposite of last June.

The Anniversary Problem

Last June ranked 123rd warmest out of 131 years in the Northeast. The heat dome pushed Boston to 102 degrees and put 160 million Americans under heat alerts simultaneously. Lowe’s July comps surged 4.7%. Management referenced weather 15 times on the earnings call.

This June is shaping up as the polar opposite — literally.

Cool and wet in the Central and Eastern US through the first three weeks of the month, measuring against one of the strongest heat signals in a generation.

The comparison is not a forecast problem. It’s a math problem.

Cool and Wet Is Not a Summer Demand Environment

In summer, driving incremental demand above the seasonal baseline requires a sustained anomaly — roughly two standard deviations from normal held over weeks, not days. A degree or two above normal is invisible to a consumer already in summer mode. It takes an outlier to change behavior.

Cool and wet is a different problem entirely. It doesn’t just fail to drive incremental demand — it actively suppresses the seasonal demand that would otherwise be there.

Low 70s in Chicago next week is not neutral for a retailer with summer seasonal product on the floor. Air conditioning units don’t move. Lawn mowers stay in the aisle. Patio furniture sits. Swimwear doesn’t convert. The consumer hasn’t changed their mind — the weather changed it for them.

The South and Southeast stay warm but face above-normal precipitation. Warm and wet suppresses the outdoor categories that warm and dry accelerates — patio, lawn and garden, outdoor living, seasonal recreation. The heat is there. The conditions to use it aren’t.

Two different problems. One consistent outcome. June is not delivering the demand environment that seasonal categories need.

The Forecasting Advantage Nobody Is Using

The weather forecast is the most accurate forward-looking signal available to any retailer with seasonal product categories. The (free!) 6-10-day, 8-14-day, and 30-day outlooks from NOAA are reliable enough to inform inventory positioning, promotional timing, and category allocation decisions weeks in advance.

Economic forecasting cannot make the same claim. There is a reason someone once said that God created economists to make weather forecasters look good.

As I’ve documented across this earnings season — AutoZone, Gap, Kohl’s, Macy’s — the weather was the silent backstop in Q1 that nobody named. The warmest spring in 130 years offset a consumer carrying a 70-year sentiment low, elevated fuel costs, and exhausted tax refunds. The weather provided the floor. The macro dug the hole. The weather won.

Q2 is a different problem. A cool and wet June doesn’t just fail to provide that floor — it compounds the macro headwinds. For the first time this year, both variables are pointing in the same direction. And neither is pointing up.

The forecast said so weeks ago. The question is whether anyone was using it.

Musical Coda

© 2026 G2 Weather Intelligence. All rights reserved.

Last week, I wrote that Gap had nowhere to hide in Q1. The warmest spring in more than 130 years couldn’t save Old Navy’s dress business.

Macy’s told a different story this morning — comparable sales up 3%, the strongest first quarter performance in four years, with dresses as a standout category in the same weather window that Old Navy couldn’t convert.

That contrast tells you something important: the weather was the same for everyone. What differed was execution, assortment, and, importantly, who walked through the door.

Macy’s customer base skews middle- to upper-income. That customer is less exposed to fuel cost pressures and the collapse in sentiment that is hitting lower-income households. With the S&P 500 near record highs and wealth effects still positive for higher-income consumers, Macy’s has a structural advantage that Old Navy and Walmart don’t.

The K-shaped economy is showing up in the earnings results in real time.

One caveat worth naming: the warm spring that helped drive Q1 traffic and dress sales at Macy’s is not going to repeat in Q2 at the same intensity. Weather is less of a macro driver for Macy’s than for a mass-market value retailer — but it is not irrelevant. And the invisible tailwind that lifted Q1 results across the board is gone.

Spring raised guidance and said Q2 trends are continuing. The World Cup, Fourth of July fireworks, and MLB partnerships give Macy's event-driven demand catalysts that substitute for weather-driven demand. They'll need them — storm clouds are gathering on multiple fronts.

But the weather was in the room on this earnings call, too. Nobody introduced it.

© 2026 G2 Weather Intelligence. All rights reserved.

“It’s a hard rain’s a-gonna fall” —Bob Dylan

CNBC’s retail roundup this week called Q1 “surprisingly robust” and correctly identified three tailwinds that buoyed results — tax refunds, buy now pay later, and consumer resilience.

What it missed? The weather.

February through April 2026 was one of the warmest springs in more than 130 years. The national temperature ran more than 5 degrees above normal for the entire window.

That is not background noise. That is a demand driver — and it appears nowhere in the analyst quotes, or the earnings call commentary from the companies that benefited most.

The asymmetry is always consistent. Weather gets named when it hurts. When it helps, it becomes execution, strategy, and consumer resilience.

The Three Tailwinds That Carried Q1

Tax refunds confirmed by Target, Burlington, Ross, and Walmart. Higher-than-normal refunds arrived at exactly the right moment to offset fuel cost pressure. Buy now, pay later adoption hit new highs across all income cohorts. Consumers were stretching further than the headline results suggested.

And then there was the weather — the silent partner nobody named.

The warmest spring in 130 years pulled forward demand, multiplied shopping occasions, and moved seasonal inventory at full price across every major retail footprint simultaneously.

All three are gone.

The Q2 Weather Signal — A Subtle But Important Distinction

Spring and summer move consumer demand differently. In spring, a single warm weekend moves the needle. Summer requires more persistent anomalies sustained over weeks, not days.

There is a second factor worth naming: climate change acclimation.

Summer demand is less sensitive to modest temperature anomalies from normal than spring demand. Consumers are already in summer mode — a degree or two above normal doesn’t change behavior the way a warm April weekend does. What moves the needle in summer is an outlier sustained over weeks, not days.

Last June and July were exactly that.

The heat dome pushed Boston to 102 degrees and put 160 million Americans under heat alerts. At Lowe’s, June comps edged up 0.3% before July surged 4.7% as outdoor, lawn, and cooling categories caught up simultaneously.

Management said it plainly: ‘Our performance clearly improved as the weather turned more favorable in June and July.’ They referenced the weather 15 times on the call.

June 2026 shows a modest warm lean in the Northeast and Mid-Atlantic — but nothing approaching last June's historic heat. The Central Plains and South sit in Equal Chances. Warm but wet in the Southeast. The threshold that was cleared decisively last June and July is not in the current forecast.

What the Maps Actually Show

June 2025: Every region above normal. Northeast 123rd warmest out of 131 years. No cold pocket anywhere in the continental US.

June 2026: Above normal in the West with a strong signal — consistent with last year, no year-on-year upside. The Northeast and Mid-Atlantic show a modest warm lean, but nothing approaching last June’s historic heat. Equal Chances across the Central Plains and South. Warm and wet in the Southeast. A below-normal temperature signal is emerging in the Central Plains and South in weeks 3-4.

The Northeast warm lean is real but modest. Leaning warm against a near-record baseline is not a tailwind. It is a smaller headwind than Equal Chances would be, but it’s not a signal that drives significant incremental demand.

The Category Risk

The single biggest weather-driven demand risk in Q2 is the cooling category. Not because June 2026 is forecast to be cold — it isn’t. But because air conditioners, fans, and portable cooling units rode one of the strongest heat signals in a generation last June. June 2026 sits in Equal Chances in the Northeast. A category that surged on last June’s historic heat is now measuring that performance against a normal year.

AutoZone named it directly on their earnings call last week. Comps decelerated to 1.3% in the final two weeks of their quarter. They are planning for a normal, if not hotter-than-normal, summer.

It’s a bad bet.

Outdoor living, patio, lawn and garden, and seasonal apparel face the same comparison problem — not because June will be cold, but because the second half of last June was historically hot, and this June (and July) offer no comparable signal.

The Macro Overlay

The CNBC piece confirms what earnings calls have been signaling all week. Walmart, Ross, and TJX guided conservatively on Q2 — not because their businesses are broken but because they (wisely) can see the tax refund tailwind fading and fuel cost pressure building. They’re also pretty weather-savvy, but that goes unmentioned (see above).

While retailers can at times be “more cautious in their guidance than the reality might suggest,” executives and analysts generally agree they could see a more strained consumer in the current quarter and the rest of the year, said Saunders.

″[That] tells you that retailers are kind of seeing the signs that some of this trough around the growth rate won’t persist across the balance of this year,” said Saunders. “Not that it will be terrible, but just the heat will come out of some of that momentum, and I think that is related to the fading impact of tax [refunds] and the picture of inflation that will probably pick up across the balance of this year.”

—Source: CNBC

Walmart’s CFO said it directly: higher tax refunds muted the fuel cost pressure in Q1. As those refunds fade, consumers will feel more of that pressure.

Layer the weather comparison on top, and the Q2 picture comes into focus.

The tailwinds that carried Q1 have all reversed or run out of steam. What replaced them is a consumer under historic financial pressure, a June weather setup that offers no national heat consensus, and a year-over-year comparison that is among the most challenging in recent memory for weather-sensitive categories.

No Backstop in Q2

CNBC called Q1 surprisingly robust. It was. Tax refunds, buy-now, pay-later, and a record-warm spring carried retailers further than the underlying consumer fundamentals suggested was possible.

Two of those three were named on virtually every earnings call. The third — the warmest spring in more than 130 years — was the silent partner nobody acknowledged.

Q2 has no equivalent tailwinds. The weather was the silent partner in Q1. In Q2, it becomes a headwind — not because summer will be cold, but because last summer was one of the hottest on record, and this summer offers nothing to match it.

The companies that guided conservatively — Walmart, Ross, TJX — are reading the signals correctly.

The weather was the backstop in Q1. There is no backstop in Q2.

Musical Coda

© 2026 G2 Weather Intelligence. All rights reserved.

Last November, I warned that Gap was right for the wrong reasons — that great weather, not execution, was supporting their results, and that they would be blindsided when conditions reversed.

What happened next made the story worse than I anticipated.

The weather didn’t reverse in Q1 2026. It got better. February through April delivered one of the warmest springs in more than 130 years. Multiple regions ranked among the warmest on record. The national temperature ran more than 5 degrees above normal for the entire window.

For Gap, and Old Navy in particular, that’s not a subtle tailwind.

It’s the closest thing to a guaranteed demand environment that a mass-market seasonal apparel retailer can get. When temperatures run 5 degrees above normal for an entire quarter, consumers transition out of the winter mindset earlier, warm weekends multiply shopping occasions, and seasonal inventory moves at full price.

Dresses, swim, and shorts — Old Navy’s core seasonal categories — should have been flying off the racks. They weren’t.

Gap’s stock fell 17% today, erasing approximately $1.6 billion in market cap. Turns out the emperor had no clothes (denim or otherwise) — and the warmest spring in more than 130 years made sure everyone could see it.

CEO Richard Dickson was direct on the call: “Bluntly, we have not had the right fashion and value equation for that category.”

He’s correct. The warmest spring in more than 130 years could not move Old Navy’s seasonal assortment. The tailwind was there. The product wasn’t.

Consider Kohl’s, serving the same customer in the same weather window. Spring seasonal up mid-teens. Same weather. Right product. Right place.

The weather removed the most convenient external excuse. What remained was entirely internal — the wrong fashion, the wrong price, the wrong value equation.

The more uncomfortable question is whether Gap has been misreading its own results for two consecutive quarters — attributing to execution what the weather was quietly delivering, and walking into Q1 with misplaced confidence.

The warmest spring in more than 130 years stripped away the last layer of cover. The forest — an assortment problem that had been there all along — became impossible to ignore.

In November, I warned that Gap would be blindsided when the weather reversed. The weather cooperated in Q1. That turned out to be the harder problem.

© 2026 G2 Weather Intelligence. All rights reserved.

I don’t normally publish two days in a row. But AutoZone’s earnings miss landed squarely in my wheelhouse, my latest podcast dropped today with eerily good timing, and it’s raining again — so my plan to mow the front yard and back forty will have to wait.

AutoZone beat Wall Street’s earnings estimate by nearly two dollars a share on Tuesday. The stock fell 9% anyway — its worst single day in four years.

The explanation was the weather.

CEO Philip Daniele was direct on the call: “This slowdown in sales was caused by unseasonably cool weather impacting our heat-related categories, which normally begin to ramp this time of year as summer heat begins to take hold.”

He named the categories — air conditioning, starting, and charging. He said the correlation between cool, wet markets and comp deceleration was visible in the data. This was not a hedge. It was the explanation.

Here is the weather data supporting his explanation,

Ten consecutive weeks of above-normal temperatures. Two weeks of below-normal temperatures—the heat categories never got the signal they needed to ramp.

Daniele confirmed it precisely. The weather data and the comp data tell the same story.

Regarding our plus 4.1%, quarterly domestic same store sales, the cadence was 5% in our first 4 weeks, 4.5% in our second 4 weeks and 2.9% over the last 4-week period of the quarter.

Now, let me address the last 2 weeks a little more specifically.

Those 2 weeks were softer than the rest of the quarter with comps of 1.3%. This slow down in sales was caused by unseasonably cool weather impacting our heat related categories which normally begin to ramp this time of year as summer heat begins to take hold. This affected both DIY and commercial.

The Summer Setup for AutoZone

Daniele told analysts he expects “a normal, if not hotter than normal, summer based on the prognostication of all the weather geniuses out there.”

I’ll leave the genius designation to others.

But I’m a firm believer in the rhythm method of meteorology (aka mean reversion) — record warm summers are almost always followed by cooler ones. Last summer ranked 120th warmest out of 131 years nationally. The Northeast hit 119th.

Normal after a near-record summer is not a neutral outcome for heat-related categories. It’s a headwind.

There’s a second risk worth watching later in the summer.

A strengthening (Super?) El Niño signal in the Pacific — still weak at this point but trending — historically produces a wetter-than-normal pattern across the Northeast and mid-Atlantic in July and August.

Warm and wet does not sell air conditioners, batteries, and cooling parts the way warm and dry does. If the El Niño signal strengthens through summer as some models suggest, AutoZone’s heat category recovery in the back half of Q4 faces an additional headwind beyond the comparison alone.

And then there's gas price inflation, consumer uncertainty, etc., etc. Based on yesterday's stock performance, maybe the 'market geniuses' know something management doesn't.

Podcasting!

One more thing before I go.

Before AutoZone reported yesterday, before the stock dropped 9%, before Daniele blamed the weather on a national earnings call, I recorded a podcast on exactly this topic. The timing is accidental. The thesis isn’t.

Content Kingdom is hosted by David G. Ewing, a weekly podcast that brings together CMOs, tech innovators, and creative leaders shaping the future of brand and content. David wanted to talk about how the weather controls buying decisions. Given what happened yesterday on the AutoZone earnings call, the conversation aged well.

Come for the weather talk, stay for the war stories …

Here’s a sneak preview. More on this later.

© 2026 G2 Weather Intelligence. All rights reserved.