From Quirk’s New York: Kurt Knapton on 20 years of online sample and what comes next
Rep Data Team · August 10, 2026 · 4 min read
At Quirk’s New York last month, Rep Data Board Member Kurt Knapton joined our EVP and Head of Research, Steven Snell, PhD, to talk about how online sample has changed over the past two decades and what that history can tell us about the data quality challenges researchers face today.
As former CEO of e-Rewards and Research Now, Kurt helped shape the early online sample industry. He was an early innovator of permission-based online panels and helped commercialize them at global scale as research moved from telephone interviewing to web-based data collection. Today, as a member of Rep Data’s board, he is applying that experience to the quality questions emerging as AI changes how research is conducted and how respondents participate.
His new eBook, From Access to Trust: What Twenty Years of Online Sample Teaches Us About Trustworthy Research, looks at how respondent quality has evolved alongside online sample, from the early panel era through marketplaces, fraud prevention and AI. During our fireside chat at Quirk’s, Kurt shared some of that history, including how the definition of respondent quality has changed and what researchers should be paying attention to next.
From finding enough respondents to knowing who to trust
Much of Kurt and Steve’s conversation at Quirk’s looked back at how online sample has changed since Kurt and his colleagues were building e-Rewards in the early 2000s. At the time, one of the industry’s biggest challenges was finding enough of the right people willing to participate in online research, and quality centered heavily on recruitment, representation and response rates. Providers knew how respondents entered their panels, maintained profiles over time and managed their participation.
Kurt talked about how that model changed as online research grew and marketplaces, exchanges and programmatic sample dramatically expanded respondent access. Researchers could reach more people, support larger studies and find harder-to-reach audiences, while respondent sourcing became more interconnected and harder for any one provider to see in full. It is a shift Kurt also explores in his From Access to Trust eBook, summed up by two questions. Twenty years ago, researchers were asking, “Can we find enough respondents?” Today, “Can we trust the respondents we find?”
The Quirk’s discussion then moved into what it takes to answer that question today. Kurt and Steve talked about sourcing and identity information alongside device intelligence, digital footprints, duplication checks and behavioral signals collected during the survey. Rather than looking to one indicator to determine quality, researchers can draw on multiple signals across the respondent experience to understand who is participating, how they are participating and whether the resulting data can be trusted.
AI brings a new quality challenge
From there, Kurt and Steve turned to AI and what it could mean for the next phase of respondent quality. Kurt talked about opportunities across the research process, including survey writing and programming, open-end coding, analysis and fraud detection. On the respondent side, however, AI is creating new quality concerns. Kurt pointed specifically to agentic AI, where agents can complete surveys on behalf of people and potentially do so at scale while appearing to be genuine human participants.
Kurt and Steve also discussed where AI-assisted and synthetic participation becomes more complicated. An AI agent might know that someone likes history and movies and use that information to approximate their preferences, but if that person hasn’t seen a particular movie, the agent can’t provide their actual reaction to it. For research based on product use, media consumption or other experiences, there is still a need to hear from people who actually had those experiences.
As researchers experiment with synthetic data, AI-assisted participation and AI-generated analysis, reliable human data also provides a basis for evaluating those outputs. As Kurt writes in the eBook, “Human benchmark data becomes more valuable—not less valuable—as AI becomes more capable.”
Kurt encouraged researchers at Quirk’s to ask where their data came from and understand how respondents were sourced, what validation was performed, which quality problems were found and how respondents were filtered. He also emphasized the importance of transparency from research partners and understanding the quality processes behind the data used to make decisions.
How to Learn More
Kurt goes deeper into the history behind his Quirk’s discussion in his ebook, From Access to Trust: What Twenty Years of Online Sample Teaches Us About Trustworthy Research, covering the evolution from proprietary panels through marketplaces and modern fraud prevention to the quality questions emerging with AI.
On August 26, Kurt will join Insight Platforms for a webinar focused on where online sample and respondent quality go from here, including the new challenges AI and agentic participation are creating for researchers.