The one question it answers
HumanCheck answers a single question: is there a real, unique, live human on the other end of this response, right now? It does not try to learn who someone is. It focuses only on confirming that a genuine, individual person is present. A respondent points their camera at their face for a brief scan, a few seconds in the browser, with no documents and no government ID.The two things it proves
Liveness
A real person is present in the moment, which is what makes bots and fake media hard to get through. The check looks for the small signals that only a living person produces right then, and is built to reject photos, recordings, a screen showing a screen, and AI generated faces.
Uniqueness
One human, one response. This keeps a single person from quietly taking the same study many times under different accounts. When a genuine respondent returns, the check can recognize them, while someone you have removed for cause finds it hard to come back under a new identity.
The return journey
Verification is not only a one-time gate. It works across visits, so you can be strict where it matters and light where it does not.1
The first check
The initial check takes seconds and produces a verdict from the face scan: verified or denied.
2
Return visits, at your assurance level
On later visits you choose how strict to be. Require a fresh scan every time, or quietly recognize a returning respondent by device, so a genuine repeat participant is not made to prove themselves again.
3
Recognized respondents and guests
A respondent who confirmed an email or phone during the check can be recognized across devices. Guests, who did not confirm either, are recognized more lightly, and their records fade over time.
What it is not
Not KYC
It does not verify a legal identity or a name, on purpose. For survey work you almost never need to know who someone legally is, only that they are a real person you have not already counted.
Beyond two-factor
Two-factor proves that someone holds a credential, like a phone or an email. HumanCheck ties the check to the actual person in front of the camera, not to a thing they possess.
Not passive bot detection
It asks for a live presence, which is a stronger signal than passive behavioral analysis.
Your respondents’ data
The scan becomes a face vector, a one-way numerical representation of the face’s geometry, used only to recognize a returning person. It is not a stored image. Your panel and your platform never touch any of that. What you receive is a verdict, verified or denied, and the reference id you chose to attach, never a face and never a name.Where it fits in a study
Put the check right before the incentive. Real people push through because they want to finish, and the accounts that balk are usually the ones you were about to pay for nothing. It sits alongside the checks you already run, your attention checks and your device signals, and answers the real-versus-fake question those were not built to settle.How the check works
Liveness and uniqueness in a few seconds, and the honest boundary.
The respondent experience
The real screens a respondent sees, from click to complete, and the demo.