Customer Verification and Counter Fraud Assessment for Branch, Contact Centre and Back-Office TeamsNobody fails a check they had time for. The question is the one with nine people waiting.

Methodology: Thirty-five original exercises across five formats: sixteen matched verification calls in eight pairs, six select-every-that-applies exercises, six binary claims, four most-and-least forced choices and three matching exercises. The declared response instruction is behavioural tendency throughout - what would you do - and one instruction covers the whole form. Construct statement: it measures what a stated pressure costs somebody's verification decisions, what they take a check to have established, whether they can tell a request that is being directed by somebody else from an ordinary one, and when they stop and refer. It does not measure knowledge of any country's financial-crime rules, any named fraud system or any organisation's own procedure, and it is not an integrity screen: a low result is a reason to look at how a desk is staffed, not a conclusion about anybody's honesty. Scoring is a matched-trial within-person contrast. The eight pairs are matched on everything the design can match - the same situation, the same four options word for word, the same key, the same response set - and differ only in whether somebody is waiting. The headline is the DIFFERENCE between accuracy on the pressured set and accuracy on the settled set, taken on the raw scale. That matters and it is easy to get wrong: chance-correcting each set against its own prior and subtracting the two corrected figures makes the difference move with the respondent's level, because the two sets have different nulls by construction and that difference is precisely the ordinary cost being measured. The correction belongs on the comparison instead, so the report prints what a respondent answering at the declared priors loses across the same two sets, and the reader sees their own difference against it. Raw accuracy is reported separately and never fused with the difference, because somebody accurate everywhere and somebody inaccurate everywhere both land near zero on a cost and those are two different people. The keys vary by design: three pairs key proceeding, three key a further check and two key declining and referring, so neither always checking nor never checking is a strategy - and on the three that key proceeding, over-verification is the scored error, because a customer refused after passing is a cost the instrument refuses to pretend is free. Half the pairs are delivered settled first and half pressured first, and the two members of every pair sit at least ten exercises apart, so order and fatigue are not confounded with the load. Every figure carries a band computed from the two sets' own binomial error, because eight trials a side cannot support a point estimate and a report that prints one is claiming precision it does not have. Constructs and sources: the experimental literature on time pressure and decision quality, in which accuracy under pressure falls through a shift to fewer cues rather than through slower processing (Payne, Bettman and Johnson on adaptive strategy selection; Maule and Edland on time pressure and choice); the vigilance and signal-detection work showing that criterion, not sensitivity, moves first under load; the social-engineering literature on authority, urgency and scarcity as compliance levers (Cialdini's influence principles, applied to fraud rather than to sales); published typologies of authorised push payment and impersonation fraud, in which the customer passes every identity check because the customer is genuinely making the request; the human-factors finding that an exception granted under pressure becomes the team's remembered precedent; work on the costs of over-verification and customer friction, which is why three of the eight pairs key proceeding; McDaniel and colleagues on behavioural-tendency response instructions; and Haladyna, Downing and Rodriguez on plausible distractors and cue control. All exercises are original works; no commercial instrument's items or name is reproduced, no bank, vendor or scheme is named, and no affiliation is implied.