Before Chris Kelly tried to measure brand advertising, he tried to move college students' textbooks and consumer goods. One experiment worked a little. The other failed. He says so plainly. The ventures belonged to a habit that began even earlier, with a web design business he started with a friend in high school: find a slow exchange, put software in the middle, and see whether the result becomes useful.
The habit survived the false starts. After the University of Notre Dame, where he studied finance and economics, Kelly went to McKinsey. He became a buyer of market and consumer research, and the experience left a mark. The products were slow, expensive and stodgy. Later, as an associate at Matrix Partners, he watched investors ask whether a prospective product had real demand. The same gap appeared from another angle. A timely answer about actual people could change a decision. The tools for getting that answer felt stuck.
Kelly moved through marketing and business development at the early fintech company SavvyMoney, then returned to building. In 2012, he and Aaron Wenger, a lifelong friend and fellow Notre Dame alumnus, started Survata. They entered Y Combinator's summer batch with a compact bargain for the internet: instead of paying to pass a publisher's wall, a reader could answer a short, anonymous survey. Brands got research. Publishers got a new stream of revenue. Readers traded about 20 seconds rather than a credit card number.
The problem beneath the product
The first version was easy to explain because the transaction was visible. A customer bought responses. Survata distributed a survey through participating publishers. The publisher shared in the fee. When Kelly spoke about Y Combinator a few years later, the lasting lesson was focus. The team considered research services and white-label tools, then returned to the core model: clients had to pay for the thing they were building.
Money followed. Survata announced $1.5 million in seed funding in 2013 and a $6 million Series A in 2015. A reported $14 million Series B arrived in 2018. The more consequential change was happening inside the product. Survey responses were becoming one small ingredient in a larger analytics system. The company connected to advertising platforms, tracked campaigns across screens and built machine-learning models to interpret what happened.
The fundraising chronology can make the company look tidier in retrospect than any startup feels from the inside. Kelly had already accumulated a useful catalog of things that did not scale: the college delivery idea, abandoned directions at Survata, and the temptation to serve every adjacent research need. His descriptions of those episodes are matter-of-fact. A failed service is called failed. A merely decent textbook exchange stays “mildly successful.” That verbal restraint makes his operating philosophy easier to trust. The lesson is carried by the sequence, not polished onto it afterward: narrow the model, collect evidence, keep the commitments that matter.
“Do what you say you'll do.”Chris Kelly on the say/do ratio
That line is Kelly's favorite piece of C-suite advice. Its charm is the absence of ornament. He calls the accumulated record a “say/do ratio,” admits his own is imperfect and treats improvement as the point. The rule also describes the business he chose. Advertising is a collection of promises: reach these people, change this perception, create demand. Measurement is the uncomfortable follow-up. Did the promise survive contact with a real campaign?
When the old name became a rounding error
The name Survata had been chosen quickly, when Kelly and Wenger were two people in an unfurnished apartment with no customers and no product. It fit the survey marketplace. By 2020, it no longer fit the company. Kelly wrote that survey data had fallen to well under 0.1 percent of the data held in its databases. The system was ingesting tens of billions of anonymized data points, tracking billions of ad impressions each month and generating automated campaign insights.
Upwave signaled motion, analytics and the desired direction of a brand chart. The rebrand also clarified the mission. Kelly argued that better proof for brand marketing could protect the media that such advertising funds, including journalism, music and video. It was a financial case with a civic aftertaste: if software could help marketers justify patient brand investment, the value would travel beyond the dashboard.
The company now describes itself as a brand outcomes measurement platform. It says it has measured more than one trillion ad impressions for more than 1,000 advertisers. Its channels span digital, connected TV, social, linear television, retail media and streaming audio. The original insight remains visible, but the unit of value has changed. A response was useful. A decision is better.
Three layers, in the right order
Kelly's most practical framework separates delivery, audience and outcome. Delivery asks whether the advertisement appeared. Audience asks whether it reached the intended people. Outcome asks whether those people changed in a way the advertiser cared about. The first two matter, he says, but they are table stakes. The energy belongs in outcomes because outcomes are why the ad existed.
Delivery
Did the ad run as planned?
Audience
Did it reach the intended people?
Outcome
What changed because it ran?
This is where Kelly's argument becomes pointed. The platform that takes the advertising money often provides the measurement too. He likens the arrangement to grading your own homework. Fragmentation makes the problem harder. A marketer buying across several platforms can receive several internally tidy reports and still lack a coherent view of incremental impact.
Incrementality is the crucial word. Attribution can assign credit to the touchpoints near a conversion. An experiment asks a different question: would the outcome have happened without the ad? Kelly's preference for the second question is consistent with the founder who once disliked research because it was cumbersome and late. Precision is valuable only when it isolates something that can guide a choice.
“Outcomes are where all the energy should be going.”Chris Kelly on advertising measurement
A report card that arrives before class ends
Traditional brand studies often read like postmortems. The campaign ends, a chart appears, and the result receives a grade. Kelly wants measurement to become an action plan. Upwave has developed in-flight indicators and predictive tools designed to reveal where a campaign may outperform before the usual statistical finish line. The ambition is operational: move money, refine creative or reconsider an audience while there is time left on the clock.
The approach also explains Kelly's current interest in AI. In 2025, he appeared at the NewFronts in a cape as Upwave cast its product as a sidekick for marketers. Under the costume was a sober product idea: use AI to summarize campaign performance and recommend optimizations. In 2026, Kelly extended the argument to agentic systems capable of making thousands of small decisions in real time. The advantage, in his telling, belongs to organizations that move and learn faster.
He couples that enthusiasm with a boundary. Measurement must remain consumer-friendly. Kelly says Upwave has worked on privacy-conscious micro-cohorts since 2020, anticipating a world with fewer user-level identifiers. His formulation is blunt: if marketing analytics is not consumer-friendly, it is not sustainable. The future he describes is more automated, but not dependent on following an identifiable person across every screen.
Kelly applies a similar revision to where work happens. Before the pandemic, the idea that everyone would not sit in an office five days a week felt alien to him. He changed his mind. He still values the physical moments that carry unusual bandwidth - a customer conversation, a negotiation, a whiteboard, a drink that turns colleagues into people rather than tiles. But he no longer treats a long commute and a desk as the default proof of seriousness. It is another measurement problem in miniature. Presence is visible. Useful work is the outcome. Confusing the first for the second produces a comforting metric and a weaker decision.
The Umbro test
For all the systems language, Kelly's own brand story is sensory. He grew up playing soccer, and the first brand he remembers loving was Umbro. Each fall brought new shorts, shirts and cleats. He remembers their smell and the social lift that came with wearing them among seven-year-old classmates. It is a useful scene because it captures what the charts are trying to detect: a brand becoming memory, feeling and preference.
The finance graduate in Kelly wants that effect measured. The former consultant wants the answer in time to be used. The former investor wants evidence of demand. The founder understands that the evidence must become a product someone will buy. None of those perspectives eliminates the small child opening a box of soccer gear. They give the experience a fighting chance in a budget meeting.
Kelly recommends two books to marketers: Made to Stick for memorable communication and The Black Swan for statistical reasoning. The pairing is almost a miniature Upwave thesis. A brand needs an idea that stays with people and an honest method for deciding whether it did. Creativity supplies the possibility. Measurement supplies the discipline.
His public work has widened with the product. He sits on the Ad Council's Leadership Council and has spoken at gatherings including Cannes Lions, Advertising Week, Beet Retreat and the Advertising Research Foundation. In 2025, he argued that media networks would spread beyond retailers into travel, finance, software and other businesses with valuable customer data. The prediction fits his older pattern: when a useful asset sits trapped inside one business model, somebody will build a transaction around it. The warning is implicit too. A new stream of advertising inventory still has to prove what it delivered.
Fourteen years after the surveywall, Kelly is still working on the same exchange. A company spends money to put an idea into the world. People encounter it amid everything else competing for attention. Somewhere between exposure and memory, something may change. His job is to make that change visible soon enough to matter.