People assume the technology is the hard part of a lie detector test. It is not — the questions are. A well-written set produces a result you can act on; a vague one produces an argument.
Here is how question sets are built, what makes a question testable, and where the limits are.
Every question must be answerable true or false by the person being tested, about a fact they would know for certain, within a defined time frame. No compound questions, no opinions, no hypotheticals, no legal conclusions.
That discipline is what stops a test drifting into interpretation. "Have you been unfaithful?" is unusable, because two people rarely mean the same thing by it. "Since January 2026, have you had physical sexual contact with anyone other than your partner?" is testable.
In relationship cases, questions tend to cover physical contact, continued contact after a stated cut-off, dating apps, and money spent outside the relationship. In workplace cases, they cover whether the person took a specific item or sum, whether they knowingly helped someone else do so, and whether they have given a truthful account in the internal investigation.
In family and recovery cases, questions cover use of a named substance within a stated period, contact with a named individual, or whether an account given to the family is complete.
Feelings, intentions, motives and open-ended "why" questions are outside the scope of credibility testing. So is anything the examinee could answer honestly while being mistaken — such as what someone else was thinking or doing out of their sight.
We also decline question sets designed to humiliate rather than resolve, and questions about protected matters that have no bearing on the issue being tested.
Before the examination your examiner spends time with you privately to draft and refine the questions, usually settling on three or four issues that actually matter. The examinee then sees the exact wording before the test begins — surprise questions are not used, because a question the examinee misunderstands produces unreliable data.
On an EyeDetect® test the agreed statements are presented on screen in a structured, repeated order and answered with a button press. That repetition is part of how the scoring model works, so the set has to be fixed before you start.
Most people arrive with something broad: "has he been unfaithful", "is she stealing from the till". Those are the starting point, not the question. The examiner's job in the pre-test discussion is to turn each one into a specific fact with a date boundary, because that is what can be tested reliably.
"Have you been unfaithful?" invites an argument about definitions. "Since January this year, have you had sexual contact with anyone other than me?" does not. The narrowing process usually surfaces disagreements about what actually counts, and having that argument before the test rather than after is one of the quietly useful parts of the day.
Where a test has been requested by a partner, an employer or a solicitor, the person requesting it usually brings the concerns and the person being tested has a say in the wording. Both need to accept each question before it is used. If someone objects to a question, the objection is worked through — often it reveals genuine ambiguity rather than avoidance.
Nothing is added afterwards. The set that was agreed is the set that runs, and the written report lists each question exactly as asked alongside the outcome, so anyone reading it later can see precisely what was and was not covered.
A single test typically covers three or four distinct relevant issues. Trying to cover more dilutes the result — extra issues are better handled in a second test.
Yes. The exact wording is reviewed and agreed before the test starts. Surprise questions produce unreliable data.
You bring the issues you need answered and your examiner turns them into testable wording with you. The final set has to meet the technical rules to be scorable.
Yes, as long as it is agreed by everyone involved before the examination starts and it meets the same rules — closed, factual and about a specific past event.
A focused set of three or four relevant questions gives the clearest result. Longer lists tend to dilute the test rather than reveal more, because broad questions are harder to score cleanly.
Services · Pricing · How it works · Book a test · Contact