Six pounds, and eleven minutes
The till rings the item at £46. The customer says the shelf said £40. He isn't angry — he's mildly surprised, and he mentions it the way people do when they assume it'll be sorted in ten seconds.
The associate says she'll check. She calls a supervisor. The supervisor walks to aisle four. Two minutes pass. The queue behind grows to five. The supervisor comes back and says the label was for the previous size, which is on the shelf below, and the customer misread it.
Maybe he did. It took eleven minutes to establish, five people watched it happen, and the difference was six pounds.
The store won and the interaction was a total loss.
The only scenario where you're provably wrong
Across three series of these posts, staff have been refusing, explaining, escalating, containing and recovering. In nearly every case the fault is arguable, absent, or the customer's.
Here the business is frequently wrong and the evidence is physically present in the building. A shelf label is a claim the store made, printed, and left where the customer could read it. That produces a dynamic nothing else has.
The customer's position is checkable, which makes doubting it expensive. Investigating implicitly says I think you may have misread or be trying it on. Occasionally true. Almost never worth what the investigation costs — because the investigation happens in public, at a till, with a queue, over an amount that is usually small.
And the maths is nobody's job. No associate has ever been shown the arithmetic: six pounds against eleven minutes of two employees' time, five customers watching, and one shopper who will remember being disbelieved. The price difference is almost always smaller than the cost of establishing who's right. Nobody says that out loud, so associates default to accuracy over economics.
That's the core of retail pricing error handling, and it's a commercial insight before it's a service one.
What to do at the till
1. Believe them, visibly and immediately. "Ah — then that's what you pay. Let me fix it." No hedging, no "let me just check." The default answer to a shelf-price claim inside a sensible threshold is yes.
2. Never send them back to look. "Could you go and grab the label?" is a small humiliation delivered in front of a queue, and it asks the customer to do the store's work to prove the store's error.
3. Honour it inside your authority; escalate instantly outside it. As with returns, the failure is the middle — the long check that ends in the same outcome.
4. Fix the label before the next customer. This is the step everyone skips. The interaction ends, the till is corrected, and the wrong label stays on the shelf for the next eleven people. The customer in front of you is a symptom; the label is the fault. Someone should be walking to aisle four afterwards, not during.
5. If you genuinely can't honour it, say what happens next with a time attached. "I can't change it at this till, but I'll have someone check the shelf now and if it says forty I'll call you today and refund the difference." A specific commitment beats a refusal every time.
6. Know where the legal floor sits. Pricing obligations for displayed prices vary by jurisdiction and some are stricter than most associates realise. Retailers should give staff a clear, local answer rather than leaving it to instinct. (Worth confirming the position in your markets rather than relying on a general rule.)
The threshold nobody sets
Here's the operational gap underneath this scenario.
Almost every retailer expects associates to use judgment on price discrepancies, and almost none give them a number. So the decision is made case by case, differently by different people, on the same £6 discrepancy — which produces exactly the inconsistency customers notice and compare.
A stated threshold — honour any shelf-price claim under X without checking — solves it in one line. It's the cheapest policy in this post and one of the most common omissions.
Four ways it goes wrong
The investigator launches a two-person search over six pounds, in public, with a queue.
The prove-it sender dispatches the customer back to the aisle to photograph the label.
The label-leaver honours the price, corrects the till, and leaves the wrong label on the shelf. Resolves one interaction and schedules eleven more.
The policy-hider says the system price is the price. True in the narrowest sense, sometimes untrue legally, and reliably the version that ends up online.
Why this doesn't get trained
It's treated as a till function. Price overrides live in the POS training: which key, what authorisation, how to log it. Nothing addresses the ten seconds in which the customer finds out whether they're believed.
Accuracy culture points the wrong way. Retail systems are built to prevent unauthorised discounting, so every instinct in the process is to verify. That's correct for shrink and wrong for six pounds at a till.
The label is somebody else's job. The associate at the till isn't responsible for shelf edges, so the actual defect goes unreported. This is a handoff problem showing up as a service problem.
And peer role play makes the customer concede. A colleague will accept "the system says £46" almost immediately. The scenario only works when the customer holds a factual position calmly and doesn't go away — which is exactly what real ones do, because they're right.
What point-of-sale dispute training can rehearse
A simulation can hold a customer who is correct, calm and unmoving, and vary whether they actually are — so associates practise the threshold judgment rather than the override keystroke. The counterparties, transcripts and rubric scoring come from Foretell AI; the price-authority thresholds and legal positions stay with the retailer.
Four to build:
- The correct customer, calm and factual, who will keep standing there. Tests whether the associate honours it fast.
- The misreader, who genuinely read the wrong label — testing whether the associate can resolve it without proving the customer wrong in public.
- The chancer, claiming a price that was never displayed, where the threshold logic is what protects the store.
- The queue-pressured version, same discrepancy with five people waiting, which is where the delay costs most.
Designing the module
Pass one — the fast honour. Score time to resolution and whether the associate expressed doubt or asked the customer to verify.
Pass two — the misreader. Score whether the associate resolved it without a public correction of the customer.
Pass three — the label. Score whether the associate arranged for the shelf to be checked and corrected afterwards.
Rubric on observable behavior: Time from claim to decision. Was the customer asked to prove it? Was the price honoured within authority? Was a specific commitment made where it couldn't be? Was the label flagged for correction? Was the customer contradicted in front of the queue?
The label-flagged measure is the one to watch. It's nearly always zero before training and it's the only step that prevents the same conversation happening again.
The operator case
The recurrence is the real cost. One wrong label generates a run of identical disputes until someone fixes it. Most retailers resolve the customer and never touch the cause, which means the same six-pound argument happens repeatedly at the same till.
Delay is more expensive than the discrepancy, and nobody has told the till. Two employees, eleven minutes, a queue of five, against a small sum — the arithmetic argues for honouring almost every claim inside a threshold, and the threshold is usually unstated.
There's a compliance dimension. Displayed-price obligations carry legal weight in many markets, and an associate improvising is improvising on the retailer's behalf.
And it's a trust event out of proportion to its size. Being disbelieved over a small amount is the kind of thing customers describe for years. The cost is never the six pounds.
For retail programmes, this is a good economics-of-service exercise: students consistently optimise for accuracy and consistently fail to price the delay.
Frequently asked questions
What should a retail associate do if the shelf price is lower than the till price? Honour it immediately if it's within authority and a sensible amount, without asking the customer to prove it — then make sure the shelf label is corrected afterwards so the same dispute doesn't repeat.
Should you check the shelf before honouring a price claim? Usually not during the transaction. The check costs more than the difference in most cases, it happens in public, and it signals that you doubt the customer. Check afterwards, to fix the label.
Is a retailer legally required to sell at the displayed price? It depends on the jurisdiction and the circumstances, and some markets are stricter than staff expect. Retailers should give associates a clear local answer rather than leaving it to judgment at the till.
What's the most-missed step in a pricing error? Correcting the shelf. The customer gets resolved, the label stays wrong, and the next several shoppers have the same conversation.
The short version
A shelf label is a claim the store made in writing and left where people could read it. When a customer says it said forty, the useful question isn't whether he's right — it's whether establishing that is worth more than the difference. Almost always, it isn't.
Believe him, fix the till, and then go and fix the label, which is the actual fault and the only part of this that stops it happening again tomorrow.
Foretell AI lets retailers build conversational simulations — including pricing disputes, till escalations, and threshold-judgment scenarios like the one above — with configurable customer counterparties, transcripts, recordings, and rubric-based evaluation. If the same shelf-price arguments keep recurring, we're happy to walk through how other operators have structured it.