Five Industry Debates Taking the Stage at IIEX West
There's a specific value in watching sessions live at an event rather than reading summaries afterward. It lines up with what cognitive neuroscientists call predictive processing: when you sit in the room, your brain actively predicts, tests, and synthesizes ideas in real time, turning raw debate into personal insight.
A static recap gives you someone else's conclusions, but it skips that actual process of discovery. That's especially true when the topics are controversial or unresolved. That's the case for being at IIEX West this year instead of waiting to read someone else's summary.
Five genuinely unresolved debates run through this year's agenda, each pulling from a different corner of the industry: whether synthetic respondents can be trusted, whether AI is closing the say-do gap or widening a new version of it, how companies like Ancestry, Reddit, and Eli Lilly are actually thinking about research these days, who's still answering surveys and what that means for the data, and what a researcher's job becomes once AI can do the easy parts.
None of them have a settled answer yet. Here's who's talking, what they're arguing, and why those arguments are worth watching unfold in person.
Can You Trust What Isn't Real?
Synthetic respondents are no longer a curiosity. They're a live debate, and IIEX West is putting that debate on stage instead of smoothing it over. Viewpoints CEO Leo Yeykelis and Stanford professor Byron Reeves are running synthetic persona studies in front of the audience, then debating in real time whether the results can be trusted for actual business decisions, drawing on 20,000 replications between them.
Three other teams are circling the same question from different angles, without knowing what the other two will say. NewtonX's Sascha Eder, alongside Canva's Rachel Ousley, is mapping where synthetic personas genuinely fit next to primary research and where they don't. Radius Tech's Andrew Elder and Caroline Pendry are showing how they validated AI-built personas against live survey results before putting them to work testing more than ten concepts for a global tech client. And Appinio's Maggie Vlaco, with Emilie Faget and Cyril Croutelle, is going straight at the practical end of the question: how you actually interact with a synthetic persona to get something usable out of it.
Whichever way the Yeykelis-Reeves debate breaks, it happens live. You get to judge the core arguments firsthand, while keeping the practical frameworks from the additional sessions in your back pocket to build upon back at the office.
The Say-Do Gap Won't Stay Solved
Researchers have talked about the say-do gap for decades. What's different this year is how many separate teams think they've found a new way to close it, and how differently they're going about it. Warner Bros. Discovery's Vera Chien is narrowing it with AI-generated stimuli that respondents can react to in ways a traditional survey never allowed. SocioLens's Joshua Corona, together with Paula Rosecky, will close it with surveillance footage paired against qualitative interviews, connecting observed behavior in transit systems and retail environments to the motivation behind it. Plus Media Solutions's Julie Davitz is catching the gap in the sixty seconds after content ends, when a viewer's motivation is at its highest and about to disappear, with more than 155,000 documented behaviors at what she describes as ten times industry norms.
Then there's the session that argues the entire premise has flipped. The Clorox Company's Kristen Griffith is making the case in "Everybody Uses AI. Nobody Trusts It." that the people adopting AI fastest, Gen Z especially, trust it the least. That's the say-do gap running in reverse: behavior racing ahead of belief.
Four speakers, four methods, one unresolved question, and none of them in such close proximity except on this stage in San Francisco.
The Names Behind the Slides Are the Point
A lot of conference content is theoretical by necessity. This year's agenda has an unusual amount of real company, real stakes work behind it. Ancestry's Kendra Shapiro, presenting with Voicepanel's John Provine, is walking through how her team moved past traditional surveys and interviews once those methods stopped capturing the emotional context that mattered. Reddit's Rob Gaige, alongside Attest's Sam Killip, is explaining how Reddit stays close to actual communities instead of the demographic proxies most research still defaults to. Andrew Embry, who leads Global Insights Innovation Capabilities and AI initiatives at Eli Lilly, sits down with Griffin & Skeggs's Susan Griffin to talk about what it takes to build an audacious insights function inside one of the most regulated industries in the world. And Everyday Health Group's Sarah Ryan is bringing BabyCenter's multi-modal research to bear on what might be the single biggest purchase-behavior pivot a person goes through: becoming a parent.
If you already track what Ancestry, Reddit, or Eli Lilly are doing with research, this is the one place all three explain it themselves, in person, on the record. Miss it, and you're waiting for someone else's summary of what they said.
Who's Really on the Other End of the Survey
There's a quieter thread running through this year's sessions, and it might be the most uncomfortable one. Enlightn's Adrien Vermeirsch is laying out new participant data (300 respondents) alongside firsthand fraud investigation and finding that two of three people who start a survey get terminated before finishing, with pay for the ones who stay often landing around $2.50 an hour. He's mapping out who actually remains in that pool: the genuinely engaged, the ones who adapt, and the ones who fake it. Microsoft's Natasha Gay is looking at the sample supply chain from the other side, drawing on experience from both the supplier and the brand side to show how opaque sourcing, cleaning, and AI involvement can be, even to the buyers who depend on that data being clean.
Together, these two sessions are the industry finally turning its own tools on itself. That's the kind of detail that doesn't survive a press release. If the panel isn't the population, and most researchers have never looked closely at how their "clean" dataset actually got built, you want to hear it from the people who investigated it, not from a slide someone else took a photo of.
The Researcher's Job Is Getting Harder to Fake
The easy narrative is that AI replaces researchers. Five sessions this year argue almost the opposite. Alex Hernandez-Brun, formerly an Insights Lead at Mozilla, makes the case directly: AI has made data abundant and accessible to nearly everyone, but data was never the same thing as insight, and turning "what happened" into "what does this mean for us" still requires human judgment, drawn from fields like anthropology, psychology, and behavioral economics.
Candescent's Flavia Stoian goes further, arguing that once AI clears away the manual work an AI-native research function was built to eliminate, researchers get repositioned as thought partners rather than order-takers.
SAP Concur's John Kolbinsky is building the accountability layer underneath that shift, an internal framework he calls evidence assurance, meant to keep source traceability, methodological context, and confidence distinct but connected now that stakeholders can query research directly instead of going through a researcher first. In his framing, the researcher stops being the bottleneck between a question and an answer and becomes the steward of the organization's entire evidence base instead.
Two more speakers are looking sideways on purpose: Charter Communications's Mike Carlon is mapping 450-plus author interviews onto product development, and Rocket's Maya Kantak is arguing researchers should be borrowing more from adjacent disciplines instead of guarding their own corner of the org chart.
The through-line: the parts of the job AI can't do are becoming the whole job. Show up in San Francisco knowing which half of your job that is.
Speakers, participating companies, sessions, and event details were confirmed at the time of publication. All speakers, sessions, schedules, and program details are subject to change due to circumstances beyond our control.