There is a number that market research agencies rarely discuss publicly and almost never include in their operational cost analysis. It is the cost of a re-field — the process of repeating a study because the data from the first attempt was too compromised to use.
Depending on the study size and complexity, a re-field costs between five thousand and fifty thousand dollars. That range is wide because the variables are wide: sample size, incidence rate, geographic spread, number of quotas. But the lower end of the range is a meaningful number for any agency, and the upper end is catastrophic for a mid-size one.
Industry estimates suggest that somewhere between one in five and one in four market research agencies experience at least one significant data quality event per quarter — a re-field, a partial re-field, or a study delivery delayed by a quality remediation process.
Why It Happens
- Bots and automated scripts: Complete surveys at machine speed, generating plausible-looking data across hundreds of identical profiles.
- VPN users: Respondents outside the target geography who mask their location using virtual private networks, corrupting geographic targeting.
- Duplicate respondents: The same individual completing the same study multiple times using different browsers or clearing cookies between attempts.
- Professional survey takers: Individuals who have learned to pass screeners and quality checks, providing minimum-effort responses to maximise earnings.
- AI-generated responses: Increasingly, large language models are being used to complete open-ended questions with plausible, coherent text that passes standard quality filters.
Most of these fraud types are detectable in real time — before the response is recorded, before the quota is incremented, before the data enters the dataset. The reason they are not being detected in real time is not technical. It is infrastructural: most agencies have no real-time detection layer in place.
The Real Cost Is Not the Re-Field
The direct cost of a re-field is significant. The indirect costs are larger. A client who receives compromised data and has to wait for a re-field misses their decision deadline. A product launch is delayed. A campaign planning cycle is disrupted. A board presentation is postponed.
The relationship cost of a quality failure is harder to quantify but easier to feel. A client who has experienced one quality failure from an agency is unlikely to commission a major study without explicit quality guarantees. Two quality failures rarely leads to a third engagement.
The cost of real-time fraud protection is a fraction of the cost of a single re-field. The decision to implement it is not a close call.
SoftSight — SurveyGuard detects fraud in real time, before it enters your data. softsight.io