Nobody Owns the Stack: Narrative and The Weather Company at Advertising Week 2026
Blog
•
6
Mins
Share


No one company owns the whole advertising stack anymore, and no one has to. At Advertising Week New York 2026, Narrative founder Nick Jordan and Mandy Kageff, who leads data partnerships for weather targeting at The Weather Company, showed what that looks like in a working partnership. The Weather Company brings the weather expertise. Narrative makes it work on every platform. The media buyer stays in their own tools.
Our earlier post, From Forecast to Audience, covered what we're building. This one follows the conversation from the stage: what each side brings, where the handoffs used to break, and how AI agents are turning weeks of back-and-forth into a single working session.
Key takeaways
- Weather predicts behavior. Most audience data only records what already happened.
- A good weather trigger mixes several conditions and measures them against what's normal for that place and season.
- Forecast triggers and recent-condition triggers reach people at different moments.
- Every platform handles a signal differently, so the data has to be normalized to run everywhere.
- AI agents can check whether a weather campaign will deliver before it launches, then close the loop once it runs.
Why is weather data different from other audience signals?
Weather data is predictive: it points to what people are about to do, not just what they already did. Nick laid out the contrast on stage. Social data is about what someone liked. Retail data is about what they bought. Search data is about what they typed. Weather "also helps predict people's behavior above and beyond what they've already done," he said.
It also reaches well past umbrellas and pumpkin spice lattes. Pollen counts drive allergy relief. Dry, cold air brings on dry eye. Rain sends people to delivery apps, and a clear, dry weekend gets them out to test drive a car. Almost every category moves with the weather somewhere.
And the weather people expect isn't always the weather they get.
What is a weather trigger?
A weather trigger is a rule that turns a campaign on when a set mix of weather conditions is forecast or happening in a specific place. A trigger rarely depends on one reading. Temperature, humidity, wind, barometric pressure and sky cover combine to shape what people do.
Why isn't a simple temperature rule enough?
The same weather means different things in different places. At 75°F, Floridians reach for a jacket and New Yorkers head outside. A useful trigger compares conditions with what's normal for that location and time of year, so a trigger like "unseasonably warm" fires on different conditions in Miami than in Minneapolis.
That's why raw weather data isn't enough. "You can't just go get the free weather data from the US government and do some targeting," Nick said. Turning conditions into a trigger takes expertise in how weather moves behavior, place by place.
Should a trigger use the forecast or current conditions?
Both work, for different moments. Forecast triggers reach people while they're planning for what's coming. Triggers based on recent conditions reach people reacting to what they just lived through.
Timing matters for planning too. Most campaigns are bought over months, so a weather plan has to look three to six months ahead, not just at tomorrow. Because no one knows exactly which weeks the right conditions will hit, a longer flight on the same budget gives a campaign more chances to be live when they do.
Why does weather data need to be normalized?
Weather data has to be normalized because every platform that activates it works differently. The number of social platforms, DSPs and SSPs has multiplied over the last decade, and each handles a weather signal in its own way:
- Refresh: can it take a signal that updates in real time?
- Reach: can it estimate audience size for a cookie-free signal?
- Format: what are its character limits and naming rules?
- Learning phase: how long must a campaign run before the platform starts optimizing?
Nick explained that moving data into platforms isn't the hard part. The hard part is moving it so it works everywhere without asking each platform to change how it thinks about the world. Narrative sits in the middle so a weather trigger means the same thing on every platform it runs on.
Normalization has to reach the buyer, too. A media buyer needs to apply their own planning and measurement tools to a weather audience. And delivery data needs to flow back, so the people building triggers can see what actually drove performance and make the next trigger better.
Weather also works best alongside other data. It tells you when to reach people, while first-party audiences, demographics and geo-targeting tell you who.
How is AI changing weather targeting?
AI is turning specialist expertise into something a media buyer can use directly. Nick calls the approach composable AI: no single company will hand a media buyer the whole AI stack. "They're going to want to use the model of their choice," he said. "They're going to want to use their internal tools."
That's why Narrative is building an AI marketplace where customers and partners bring their skills, tools and MCP servers to assemble an AI stack for a specific use case. For weather targeting, that stack has four layers:
- A weather expert's skills for building triggers
- Narrative's skills for normalizing and delivering data
- The APIs of the platforms where the targeting runs
- The agency, media buyer or brand, plugging it all into their own tools
What does that look like in practice?
On stage, Nick and Mandy role-played a brief: sell more patio furniture to first-time homeowners in the Northeast this spring. The weather side of the work finds the conditions that move patio furniture, like clear skies and warm, dry weekends, and sets the trigger to fire on a different date in each market as spring arrives.
From there, the agent in the Narrative platform looks at the triggers, the historical data and each platform's inventory, then reports back whether the plan will deliver. "You don't want to propose a million-dollar campaign" that has delivered $25,000 three months in, Nick said, because the weather or the inventory never showed up. Once the plan is approved, Narrative distributes the data to each platform and brings delivery data back to close the loop.
That back-and-forth used to take weeks. "If you would have asked me 18 months ago… the whole thing is going to take a month and a half to put together," Nick said. "We've seen this work end to end, literally in a single session. 20, 30, 40 minutes."
What kinds of audiences can weather reach?
Weather can reach almost any audience, including ones that don't look weather-related at first. The session closed with a lightning round that paired audiences with triggers:
- Dry eye sufferers: poor air quality, high pollen counts, and dry, cold air.
- Spontaneous quick-service diners: rain, snow or other bad weather that sends people to the drive-thru or to delivery.
- Dry shampoo users: rain and humidity, when nobody wants to restyle their hair.
- Car shoppers: clear skies and no rain, the kind of day people go out for a test drive.
The quick-service example opened a wider point. Nick noted that a delivery company could use the same trigger to get more drivers on the road before a storm. Weather triggers can run operations as well as advertising.
Put weather to work in your campaigns
Weather tells you when. Narrative makes it run on every platform you buy on. Talk to our team to learn more.
Explore related
Insights, stories, & resources for the teams building modern data infrastructure.

Blog
•
5
Mins
What Is NQL? Narrative Query Language Explained

Blog
•
3
Mins
Easily Turn Any Snowflake Table Into a Narrative Dataset



