Data from ReDem in the Fraud & Data Quality report said that the proportion of poor quality data had risen from 12% in the first quarter of 2023 to nearly 20% in the same period in 2024.

The findings are based on analysis from Ayda’s global incentive payment platform and ReDem’s in-survey data cleaning tool.

Ayda benchmarks examining the financial impact of fraud found an average of one in five incentive payments were being flagged as potentially fraudulent.

ReDem said its research detected widespread nonsense or low-effort answers, such as random keystrokes or repeated patterns, with more than 20% of responses on the ReDem platform were flagged as ‘inattentive’.

Other examples of fraudulent responses to surveys included Ayda discovering a single individual completing studies for 15 different research companies within two months, and ReDem flagging hundreds of AI-generated survey completions where machine-written responses initially appeared human.

Shifra Cook, chief executive and founder at Ayda, said: “Our data shows the scale of the problem is not theoretical, it’s happening every day.

“Fraudulent respondents, bots, and inattentive survey takers are costing the industry billions and undermining trust in research.”

Florian Kögl, chief executive at ReDem, said: “Quality is the foundation of meaningful survey data. Without reliable insights, decisions can quickly move in the wrong direction.

“As AI continues to rise, the importance of high-quality data grows even further, since this information is increasingly reused to train synthetic samples. At the same time, ensuring data quality is becoming more challenging as we work to separate authentic human responses from misleading or artificial ones. Our mission is to preserve trust in research by filtering out what is not human and not trustworthy.”