With 5 billion survey scans annually, we have extensive visibility into fraud patterns across the survey ecosystem. That visibility informs our machine learning loop, helping improve the defenses below, identify risky sources, and protect your study as fieldwork progresses.
Review technical identity, known fraud history, respondent activity, and external reputation before incoming traffic reaches the questionnaire.
Every respondent is scanned against more than 130 signals, including hardware specifications, browser data, WebRTC overlays, and fraud history. Session profiles are built in milliseconds to support pre-survey review.
A daily-updated suppression list checks respondents against known bad actors. A respondent flagged elsewhere in the ecosystem can be blocked before survey entry.
Cross-ecosystem visibility surfaces when a respondent has attempted 30, 50, or more than 100 surveys in a 24-hour period, helping identify unusual survey velocity.
Signals span IP reputation, bot detection, device intelligence, ad fraud, identity verification, proxy detection, browser integrity, and threat intelligence.
Evaluate response characteristics during or after fieldwork. Each control offers a distinct view for practical quality review.
Evaluate each open-ended answer for relevance, effort, language, AI-generated content, duplicates, nonsensical text, and copy-paste. Responses receive an individual quality score.
"I love this product because it has many features that I enjoy. The quality is great and I would recommend it to everyone."
"honestly the battery life is what sold me - i had a samsung before and it'd die by 3pm every day. this one lasts til i go to bed"
"good product nice quality"
"The ergonomic design really helps with my carpal tunnel. I switched from a standard mouse and the difference after a week was noticeable in my wrist pain."
Surface contradictions across a respondent's answers with a specific explanation for each flagged inconsistency, supporting review across question types.
Review keystroke rhythm, pauses between inputs, and typing speed to distinguish natural interaction from bot-generated or copied responses.
Compare each respondent's completion speed with question-level benchmarks. Consistently fast answers can be flagged for review.
Detect mechanical patterns in grid and matrix questions, including straightlining, diagonal patterns, random clicking, and other non-engaged behavior.
How would you rate the quality of our product and service?
| 1 | 2 | 3 | 4 | 5 | 6 | |
|---|---|---|---|---|---|---|
| Item A | ||||||
| Item B | ||||||
| Item C | ||||||
| Item D | ||||||
| Item E | ||||||
| Item F |
Use RepData's Research Platform to connect sample, fieldwork, and quality review in one workflow. Quality evaluation also supports a native Decipher connection, API-based workflows where the survey platform supports the required API, and raw data upload for completed surveys.
Explore published examples and their original study context.
Media & Publishing
Standing behind every survey result meant manually reviewing every single interview, a highly time-consuming and resource-intensive process.
10 hrs Saved on cleaning per tracking wave
Read the storyPolitical Polling
Massive amounts of bad data from polls required extensive back-end cleaning, threatening fast-paced weekly reporting deadlines.
<1% Bad data rate
Read the storyIt brings together pre-survey fraud prevention and in-survey response quality evaluation. Each addresses a different stage of the research workflow.
Pre-survey screening considers technical and behavioral signals associated with incoming traffic before a respondent enters a study.
Checks can evaluate open-ended responses, coherence across answers, behavioral patterns, response timing, and grid-question patterns.
Signals support review and study rules. Teams apply quality findings within the workflow and decisions configured for the project.
Share your survey setup, existing controls, and what you need to evaluate.
Kantar Info Research Austria
[Rep Data] helps us identify and exclude interviews that do not meet our high standards during fieldwork, in an innovative and time-efficient manner. Particularly noteworthy are the click pattern recognition in grid questions and the AI-assisted validation of open-ended responses.
Michael Bieber
Managing Director