Home / Free audit

Free tool · 3 minutes · no sign-up

Interview Cheat-Risk Audit

15 questions about how your interview loop actually runs today. You get a risk score out of 30, a plain-English diagnosis, and the three fixes that matter most for your loop. It runs in your browser. Nothing is sent anywhere.

Answer for what happens in most of your interviews, not what’s written in a doc nobody reads. If you’re torn between two answers, pick the riskier one.

Area 1

Policy

Q1 of 15Do candidates receive a written AI policy before their first technical stage?

No written policy.+2 riskA policy exists but is vague ("use good judgment"), or candidates only hear it on the call.+1 riskA written policy is sent with the invite and linked from the job post.+0 risk

Q2 of 15Is every stage labelled AI-allowed or independent, and does every interviewer know which?

No stage labels.+2 riskLabels exist, but interviewers or candidates are sometimes unclear.+1 riskEvery invite and every interviewer brief states the lane.+0 risk

Area 2

Format

Q3 of 15Does each stage have one or two written competencies it measures, and only those?

Stages are "technical interview 1, 2, 3" with no defined focus.+2 riskFocus areas exist informally, or overlap heavily between stages.+1 riskEach stage has written primary and secondary competencies, with no overlap.+0 risk

Q4 of 15Are your technical questions your own, rather than well-known puzzles with answers published online?

Mostly standard algorithm puzzles or questions from public lists.+2 riskA mix, or your own questions haven't changed in a long time.+1 riskQuestions are based on your real work and rotated regularly.+0 risk

Q5 of 15Does every technical question have written follow-ups that perturb, derive, and probe ownership?

Interviewers ask the main question and move on.+2 riskSome interviewers ask follow-ups, but they're improvised.+1 riskEvery question has a written four-step ladder (clarify, perturb, derive, own) in the brief.+0 risk

Q6 of 15Does the candidate explain and modify their own work live, in an independent stage?

No. Take-home work is scored on the submission alone, or there's no work sample.+2 riskThere's a review, but it's a presentation without live changes or probing questions.+1 riskEvery work sample is followed by a live review with "why" questions and at least one change made during the call.+0 risk

Area 3

Interviewers and scoring

Q7 of 15Can your non-technical screeners tell a strong answer from a weak one?

Recruiters screen on keywords or confidence.+2 riskRecruiters have some questions but no guidance on answers.+1 riskEvery screening question has written "listen for" cues for strong and weak answers.+0 risk

Q8 of 15Do interviewers score against anchored scales and write evidence for each score?

Thumbs up or down, or free-text impressions.+2 riskA numeric scale, but no anchors, or evidence is optional.+1 riskA 1–4 scale with written anchors, and at least one quote or action per score.+0 risk

Q9 of 15Does every interviewer submit scores before reading others' notes or discussing the candidate?

Interviewers chat about candidates between stages, or decide together in a meeting.+2 riskUsually, but not enforced.+1 riskAlways. Scores are submitted before any discussion, and the debrief follows a set structure.+0 risk

Q10 of 15Have your interviewers been trained and calibrated on your loop in the last six months?

No training. People learn by doing.+2 riskInformal shadowing, but no calibration exercise.+1 riskEveryone has done a calibration session, scoring the same answer and comparing.+0 risk

Q11 of 15Do interviewers have a written protocol for suspected hidden help?

No. Interviewers handle it however they see fit.+2 riskThere's general guidance, but no exact lines or note format.+1 riskInterviewers use a protocol: probe with follow-ups, never accuse, record observations not conclusions, escalate to a named person.+0 risk

Area 4

Verification

Q12 of 15Is identity verified for every candidate before an offer, using the same method for all?

No identity check, or only for candidates who "seem off".+2 riskA check exists but isn't consistent, or relies only on the CV and LinkedIn.+1 riskEvery candidate completes a live ID match or provider check before offer, and the person matches the one from the interviews.+0 risk

Q13 of 15Do reference checks verify specific interview claims, through independently verified contacts?

No reference checks, or a generic "would you hire them again?" call.+2 riskReferences are called, but with generic questions or using only the contact details the candidate supplied.+1 riskReferees are verified independently and asked open questions about three specific claims from the interview.+0 risk

Area 5

Access and onboarding

Q14 of 15Do new hires get system access in stages, with payroll and equipment tied to their verified identity?

Full access on day one; equipment and pay go wherever requested.+2 riskSome staging of access, or payroll is checked but equipment isn't.+1 riskProduction and sensitive access expands over the first weeks; pay goes only to the verified name, and equipment only to the verified address.+0 risk

Area 6

Validation

Q15 of 15Do you run a structured 30-day plan that checks the claims made in interviews?

No. Onboarding is unstructured and performance concerns surface months later.+2 riskOnboarding exists, but isn't linked to interview claims.+1 riskEach hire has a written plan with tasks mapped to specific claims, weekly check-ins, and a 30-day review.+0 risk

The scoring needs JavaScript. You can still score yourself: add up the risk points next to each answer, then compare with the bands below (0–4 Low, 5–11 Moderate, 12–19 High, 20–30 Critical).

About the audit

What does this audit measure?

It measures your loop, not your candidates. The question isn’t “is anyone cheating?” You can’t answer that reliably, and trying to answer it is how good candidates get wrongly rejected. The question is: if a candidate had hidden help, how far would it get them, and how soon would you find out?

The 15 questions cover six areas. Policy: do candidates know the rules before they prepare? Format: do your questions keep moving after the first answer? Interviewers and scoring: does everyone score the same way, on evidence? Verification: is the person you hire the person you interviewed? Access: what can a new hire touch on day one? Validation: do you check interview claims in the first 30 days?

Why score the loop instead of trying to catch cheaters?

Because detection is an arms race, and the hiding tools update as fast as the detectors. Overlay assistants stay invisible to screen share. Voice-mode models don’t touch the browser at all. A webcam shows you a person looking slightly off-screen, which is also what a nervous or neurodivergent candidate looks like.

Format changes don’t have that problem. A question that changes on screen halfway through, a “why this and not that?” about a decision made in the room, or a request for the details of something the candidate shipped: none of these can be answered in advance. So the audit weights the things that make hidden help less useful, rather than the things that try to spot it.

How is the score calculated?

Each question scores 0, 1 or 2 risk points: 0 when the practice is consistent for every candidate, 2 when it doesn’t happen. The total is out of 30, where higher means more exposed.

0–4, Low. Hidden help gets candidates little, and you’d catch a mismatch early.
5–11, Moderate. Your loop produces real signal. Remaining gaps are mostly verification and validation.
12–19, High. Some good practices, but gaps a prepared candidate can use, and inconsistency between interviewers.
20–30, Critical. Hidden help can carry a candidate through most of your loop.

Your top three fixes are ranked by your riskiest answers first, then by the order that matters most for your band. A loop in the Critical band should fix its format before it buys anything. A loop in the Moderate band should close the gaps after the interview.

Three questions are treated as urgent whatever your total: no live defense of the candidate’s own work, no consistent identity check, and full system access on day one. Those are the gaps where a proxy or a fraudulent hire gets furthest.

What should I do with my result?

Fix one band at a time, not everything at once. Score each role separately if your loops differ. Better still, have two people take the audit independently and compare. Any question where you disagree by more than a point is a practice that isn’t consistent yet. Re-run it after each hiring cycle; the shareable link keeps your answers.

If your first fix is the AI policy, the two-lane policy template is free. The templates, ladders and scorecards for every other fix are in the full kit.

Audit FAQ

Are my answers stored anywhere?

No. The audit runs entirely in your browser. Your answers only leave the page if you copy or share the result link, which encodes them in the part of the URL after the # sign. That part is never sent to a server.

Where do the questions come from?

They’re the cheat-risk self-audit from the Templates book of the Unscripted kit (Template 12). The template numbers mentioned in each fix refer to that book.

Does a low score mean nobody can cheat in my interviews?

No. It means hidden help gets a candidate less far, and you’d find out sooner. No format makes cheating impossible. A good one makes it expensive and short-lived.

Who should fill it in?

The hiring manager or whoever owns the loop, ideally with one interviewer who runs it day to day. Recruiters can take it too: several questions are about screening and identity checks, which often sit with recruiting.

Next step

Every fix in this audit maps to a template in the kit.

Policies, ladders, interviewer briefs, the suspicion protocol, ID checklist, reference script, debrief and 30-day plan. Ready to paste.

See the full kit Get the free policy template