Data Integrity in the Marketplace Era
In the evolving world of market research, the conversation around data quality has never been more crucial. In this episode of *The Collaborative Canvas Podcast*, Ankesh Saxena sits down with James Rogers, Managing Director - APAC at PureSpectrum, to explore one of the industry's most pressing challenges: maintaining high-quality data in a marketplace sampling model.
James, with over 18 years in the insights and data collection space, brings a clear-eyed view on what’s broken—and what’s being done to fix it. His career spans legacy panel models to the rise of digital marketplaces, giving him a front-row seat to the transformation of sample sourcing and the rise of fraud.
**The Decline of Proprietary Panels**
One core issue discussed was the slow death of proprietary panels. James emphasizes that with very few exceptions, large-scale research has moved toward multi-sourced marketplaces. But this shift brings risks: click farms, bots, ghost completes, and survey farms are common threats to data integrity.
**Enter PureScore**
How does PureSpectrum address this?
PureScore is a proprietary system developed to measure and manage respondent behavior. Every new respondent is evaluated at entry via a unique ID, scored on their interactions, and tracked continuously. Those falling below a certain score (5/10) are automatically filtered out, ensuring low-quality participants never reach the survey.
**Detecting Subtle Fraud with PureText**
The company also introduced PureText, an AI model designed to detect gibberish, check language consistency, and evaluate the validity of open-ended responses. These innovations help flag subtle forms of fraud that can’t be caught by traditional methods.
**Shared Responsibility for Quality**
James makes it clear that platforms can’t do it alone. Marketplace providers must work hand-in-hand with clients. Survey design, targeting accuracy, and built-in validation questions (like honey traps or open-ends) are vital to creating reliable datasets.
Fraud isn’t always obvious. It's often a pattern—not an event. That’s why longitudinal data is key. With billions of sessions collected over seven years, PureSpectrum is able to flag suspicious respondent behavior over time.
**Benchmarks That Matter**
One highlight of the discussion was the platform’s global reconciliation rate—just 7%. For the industry, that’s a benchmark. It means the vast majority of collected data aligns well with expected outputs, despite increasing fraud complexity.
James also reflects on conversion rates. While they vary by project, PureSpectrum sees strong results for large-scale programmatic clients. Niche studies, understandably, see lower rates due to narrower targeting, but tools like PureScore help mitigate these issues by optimizing respondent quality.
**Looking Ahead: AI and Ethics**
As AI becomes more central to the conversation, James doesn’t shy away from the trend. He anticipates more innovation through AI-powered fraud detection, digital twins, and even synthetic data. But he’s also cautious:
> "We can’t just keep going lower and faster. At the end of the day, we’re working with humans."
This conversation is a must-listen for researchers, agencies, and clients navigating the modern research landscape. From building better respondent experiences to maintaining ethical incentives and designing thoughtful surveys, every part of the chain matters.
**Final Reminder**
James leaves us with a powerful reminder:
> *Data quality is a shared responsibility. It’s not about blaming suppliers or squeezing timelines—it’s about designing systems that work for everyone in the ecosystem.*
James, with over 18 years in the insights and data collection space, brings a clear-eyed view on what’s broken—and what’s being done to fix it. His career spans legacy panel models to the rise of digital marketplaces, giving him a front-row seat to the transformation of sample sourcing and the rise of fraud.
**The Decline of Proprietary Panels**
One core issue discussed was the slow death of proprietary panels. James emphasizes that with very few exceptions, large-scale research has moved toward multi-sourced marketplaces. But this shift brings risks: click farms, bots, ghost completes, and survey farms are common threats to data integrity.
**Enter PureScore**
How does PureSpectrum address this?
PureScore is a proprietary system developed to measure and manage respondent behavior. Every new respondent is evaluated at entry via a unique ID, scored on their interactions, and tracked continuously. Those falling below a certain score (5/10) are automatically filtered out, ensuring low-quality participants never reach the survey.
**Detecting Subtle Fraud with PureText**
The company also introduced PureText, an AI model designed to detect gibberish, check language consistency, and evaluate the validity of open-ended responses. These innovations help flag subtle forms of fraud that can’t be caught by traditional methods.
**Shared Responsibility for Quality**
James makes it clear that platforms can’t do it alone. Marketplace providers must work hand-in-hand with clients. Survey design, targeting accuracy, and built-in validation questions (like honey traps or open-ends) are vital to creating reliable datasets.
Fraud isn’t always obvious. It's often a pattern—not an event. That’s why longitudinal data is key. With billions of sessions collected over seven years, PureSpectrum is able to flag suspicious respondent behavior over time.
**Benchmarks That Matter**
One highlight of the discussion was the platform’s global reconciliation rate—just 7%. For the industry, that’s a benchmark. It means the vast majority of collected data aligns well with expected outputs, despite increasing fraud complexity.
James also reflects on conversion rates. While they vary by project, PureSpectrum sees strong results for large-scale programmatic clients. Niche studies, understandably, see lower rates due to narrower targeting, but tools like PureScore help mitigate these issues by optimizing respondent quality.
**Looking Ahead: AI and Ethics**
As AI becomes more central to the conversation, James doesn’t shy away from the trend. He anticipates more innovation through AI-powered fraud detection, digital twins, and even synthetic data. But he’s also cautious:
> "We can’t just keep going lower and faster. At the end of the day, we’re working with humans."
This conversation is a must-listen for researchers, agencies, and clients navigating the modern research landscape. From building better respondent experiences to maintaining ethical incentives and designing thoughtful surveys, every part of the chain matters.
**Final Reminder**
James leaves us with a powerful reminder:
> *Data quality is a shared responsibility. It’s not about blaming suppliers or squeezing timelines—it’s about designing systems that work for everyone in the ecosystem.*
Inspired from episode: Data Integrity in the Marketplace Era
Listen to the full episode with “James Rogers” on The Collaborative Canvas Podcast—now streaming on YouTube and Spotify.
Back to All Blogs