Bots, speeders, straight-liners, duplicates. Winnow scores every survey response and flags exactly why it can't be trusted — in one API call, in milliseconds.
AI-generated answers and incentive-chasing respondents are flooding panels. One bad response skews a finding — and the decision that follows it. Most teams catch them by hand, late, if at all. Winnow checks every response the moment it arrives.
Each response gets a quality score from 0–100, a recommendation — accept, review, or reject — and the exact flags that fired. Six independent checks. No machine-learning guesswork.
Responses finished far faster than is humanly plausible.
The same option picked down an entire matrix battery.
Failed trap questions you define in the payload.
Hashed fingerprints that have already been seen.
Open-ended answers that are keyboard mash, not language.
Near-identical time spent on every single question.
Post a response with its answers and timing. Get back a score and flags you can act on instantly.
# Score a single response POST https://api.licrat.com/v1/score { "response_id": "resp_8843", "duration_seconds": 8, "survey": { "min_expected_seconds": 60 }, "answers": [ … ] } # → response { "quality_score": 22, "recommendation": "reject", "flags": [ { "code": "speeding", "severity": "high" } ] }
Deterministic rules, no language-model cost per call. Scoring runs in milliseconds.
A single endpoint you wire into any survey, panel, or platform pipeline.
You send hashed fingerprints, never raw personal data. Nothing to leak.
Every plan runs the same 6 deterministic rules — the only difference is how many responses you can score per minute.
Same detection on every plan — you only scale the volume.
Winnow is in early access. Join the waitlist and we'll reach out with an API key.