The study, outlined in the white paper Elevating Data Quality Through Collaboration: How Shared Signals and Cross-Industry Cooperation Improve Research Outcomes, found that sampling is more complex than expected, and data quality is impacted by provider choice and survey design. The study also found that the type of data cleaning used changed research conclusions.  

The research by Data Quality Co-op also found that not all poor-quality data stems from survey fraud. The researchers segmented respondents into four categories based on 13 quality signals spanning technical signals, in-survey signals and sources signals.

The four participant segments were: ‘fingerprint flagged’, which represented 11% of participants and were characterised by fraudulent technical quality indicators; ‘bad open-enders’, which comprised a quarter ( 26%) of participants and provided low-quality data; ‘professional panellists’ representing 29% of participants and ‘survey newbies’ ( 34%), which exhibited inconsistent behaviour, such as missing answers.

The report proposed that the market research industry should combine multiple quality signals from buyers, suppliers, platforms and fraud tools, which it said would build a ‘more complete and accurate’ picture of data integrity.

Bob Fawson, co-founder of the Data Quality Co-op, said: “The industry’s current approach to data quality is too reactive and siloed. This paper outlines a clear path forward that prioritises cooperation, structure, and transparency. If we want to generate reliable insights from first-party data, we need systems that support trust and repeatability from the start.”

“We need to stop thinking of quality as something we clean up after the fact,” said Ian Haynes, co-founder of Data Quality Co-op. “Quality starts upstream, and that means improving visibility into how respondents are sourced, how they behave, and how our decisions, like question routing and sample filtering, impact outcomes.”

Data Quality Co-op conducted the research in March 2025 with over 2,000 US parents of children aged 0 to 17. Respondents were sourced from two major sample providers and recruited through various digital channels, including panels, ‘get-paid-to’ sites and fintech apps. Participants completed a 12-minute survey on food pouches that collected brand preferences, product evaluations, open-end responses and self-reported sourcing information.