Why an £8 Mistake Can Cost More Than an £800 one - The Economics of Customer Forgiveness. The Full Argument
I am working from home today because Amazon failed to deliver a large-screen television.
It was meant for our new office. It was expensive. It should have arrived yesterday. It did not.
Yet I am not particularly worried.
I expect it will arrive today. If it does not, I have a fairly confident idea of what will happen next: I will be able to find the order, speak to someone or use the app, and receive either another television or my money back.
Amazon has made a fairly significant mistake and my reaction is: no problem.
Now compare that with another large company I occasionally use to buy second-hand goods.
I recently ordered something costing about £8 or £9. It was supposed to arrive within a week. Three weeks later, it has not arrived. I have received no meaningful updates and I do not know what is happening.
The financial risk is almost irrelevant. But the experience has irritated me enough to confirm something I already suspected: I will do almost everything I can to avoid using that company again.
So one company has let me down over an expensive television and I have effectively forgiven it already. Another has let me down over an £8 purchase and may have lost me as a customer.
This appears economically irrational.
It is not.
The size of a company's mistake does not determine whether a customer forgives it. What matters is what the customer believes will happen next.
Every mistake has two sizes
Companies tend to think about failure objectively.
How much money was involved? How late was the delivery? How serious was the error? How quickly did we close the complaint?
The customer experiences a second, less visible size: the amount of physical, cognitive and emotional work required to put the failure right.
A restaurant can serve an undercooked meal and recover by replacing it quickly, apologising and allowing the customer to continue their evening. Turn the same meal into an argument, a long wait and a search for the manager, and the incident acquires a second life.
The original failure belonged to the kitchen. The recovery effort was transferred to the customer.
That is why a cancelled flight can sometimes be forgiven more easily than a £20 billing error.
If an airline explains the cancellation, puts the passenger on the next flight and sends meal vouchers to their phone, the failure remains serious. But the customer can see ownership, movement and an end.
If recovering £20 requires six menu options, three conversations, repeated evidence and a fortnight of chasing, the financial mistake is small but the psychological message is enormous:
Your time does not matter to us.
Customers do not judge a failure only by what went wrong. They judge what the organisation makes them do next.
Forgiveness is a prediction about the future
We often describe trust as a belief that a company will do what it promised.
That is only half of it.
Real trust also contains a recovery belief: if this company fails, I will not be left alone with the consequences.
That is what I have with Amazon. It is not a belief in perfection. Yesterday disproved that. It is a belief in recoverability.
Amazon has deliberately reduced the perceived risk of failure. Its UK returns operation offers free options on most eligible items, with label-free and box-free drop-offs widely available. More than 95 per cent of UK customers are within 5km of a free return point. The logistics matter, but the psychology matters more: the customer knows there is a route back.
The Ritz-Carlton has designed for the same principle in a more human setting. Its service-recovery approach emphasises employee empowerment, authentic apology and personal ownership of the guest's problem. The two companies look nothing alike. One runs fulfilment networks; the other runs luxury hotels. But both understand that trust is created partly by making recovery credible before it is needed.
This gives us a better definition of forgiveness.
Forgiveness is not the customer deciding that the failure did not matter. It is the customer deciding that the failure is not reliable evidence of what the whole future relationship will feel like.
The first company made a mistake.
The second company revealed itself.
Even bees learn when not to return
There is an odd parallel in nature.
Some orchids attract pollinators without offering the nectar or other reward the insect expects. The deception can work, but pollinators are capable of learning from unrewarding visits and changing where they forage.
The orchid does not hold an exit interview. The bee simply stops coming back.
Customers behave similarly. They do not always complain. They do not always explain why they left. They accumulate evidence.
The missing update. The information repeated for the third time. The call that promised a response which never came. The chatbot that made the company easier to contact but no easier to get an answer from.
Each interaction teaches the customer something about the likely cost of the next interaction.
Eventually the customer learns to forage elsewhere.
This may be why companies systematically underestimate the commercial cost of effort. Their complaint data captures the people who protested. Their churn data arrives later. Neither necessarily captures the moment when a customer quietly decided: never again unless I have no alternative.
The cost of being unforgiven
The latest evidence suggests that these private decisions are commercially significant.
In PwC's 2025 Customer Experience Survey of 5,511 consumers, 52 per cent said they had stopped using or buying from a brand because of a bad product or service experience. Twenty-nine per cent specifically attributed leaving to poor customer experience. Yet roughly nine out of ten executives believed loyalty had increased in recent years, compared with only four out of ten consumers.
That is not simply a perception gap. It is a failure to see the evidence customers are collecting.
The insurance industry provides a particularly stark example because the moment of failure often arrives when the customer is already under stress.
UK financial firms received 1.36 million complaints about insurance and pure-protection products during 2025: 697,635 in the first half and 665,370 in the second. Firms upheld more than half of all financial-services complaints they closed in the second half of the year.
At the Financial Ombudsman Service, 32 per cent of the complaints resolved in the first half of 2025 were upheld in favour of the customer.
These figures do not prove that every complainant subsequently left. Nor does a renewal automatically prove loyalty; many policyholders stay because switching feels difficult or because the price happens to be acceptable.
But they reveal an enormous number of moments in which customers were asked to reconsider what sort of organisation they were dealing with.
And insurance renewals are economically unforgiving. Consumer Intelligence reported that only 33 per cent of motor customers and 36 per cent of home-insurance customers switched provider in the first half of 2025. That may sound reassuring. It should not. A customer who stays because of inertia is not the same as a customer who trusts you. The first can be bought by the next attractive price. The second is harder for a competitor to dislodge.
The commercial question is therefore not merely, “Did the customer renew?”
It is, “What did the claim or service failure teach them about renewing next time?”
Why delight is often the wrong remedy
There is a well-known idea in customer experience called the service-recovery paradox: under certain conditions, a customer whose problem is recovered exceptionally well can become more satisfied than someone who never experienced a problem.
It is an attractive idea because it converts failure into opportunity.
It is also easy to misuse.
Research findings on the paradox are inconsistent. It appears to depend on the severity of the failure, the customer's previous relationship with the company, the quality of communication, the perceived fairness of the recovery and the scale of any compensation. In one 2022 hotel study, compensation had to reach roughly 80 per cent of the original booking value before satisfaction surpassed the error-free benchmark.
In other words, recovery can restore trust and sometimes strengthen it. But deliberately relying on failure to create loyalty would be like burning down the restaurant to demonstrate the efficiency of the fire brigade.
The more useful lesson is simpler: because failure is inevitable, the ability to recover is part of the product.
This is also why “delight” can be an expensive distraction. A voucher after three weeks of chasing may be generous, but it does not return the customer's time. An apology delivered by the sixth person does not undo the five previous hand-offs.
Matthew Dixon, Karen Freeman and Nicholas Toman's research into service interactions found a striking asymmetry. Among customers reporting low-effort experiences, 94 per cent intended to repurchase and 88 per cent intended to increase spending. Only 1 per cent intended to speak negatively about the company. Among customers who struggled to resolve their problem, 81 per cent intended to spread negative word of mouth.
These are stated intentions, not guaranteed behaviour. But the direction is difficult to ignore.
Exceeding expectations produced only marginal loyalty gains. Making customers work hard produced disloyalty.
The customer did not need theatre. They needed the television, the refund or the claim resolved.
Customer effort comes back as operating cost
The strange thing about customer effort is that organisations do not escape it by transferring it.
It comes back.
Customers who do not know what is happening call for updates. Customers who are passed between teams repeat information. Customers who receive unclear requests send the wrong evidence. Customers whose promises are broken complain and escalate.
More contacts create more tasks. More tasks create more queues. Longer queues create more delay. More delay creates more contacts.
What managers describe as demanding customer behaviour is frequently demand created by the design of the organisation itself.
I saw this in one global insurer where complex claims were taking an average of 508 days to settle. Work moved through queues, referrals and legal hand-offs. Everyone looked busy, but the customer experienced waiting, repetition and uncertainty.
The insurer redesigned the work around clearer ownership, fewer hand-offs and faster end-to-end decisions.
Settlement time fell from 508 days to 35 days. The number of claims closed with the same headcount increased 2.8 times. Failure-demand calls fell by 61 per cent. Touches fell by 44 per cent and the payment lifecycle fell by 73 per cent.
The claims did not suddenly become simple. The insurer simply stopped asking customers to carry so much of its complexity.
That is the point leaders often miss. Making recovery easier is not merely an act of generosity towards the customer. It can reduce the organisation's own workload.
Low effort is one of those rare ideas where the interests of the customer and the company are not in conflict.
Low Effort → Loyalty → Profit
Design for forgiveness
Most organisations invest heavily in preventing failure. They should.
But no company can eliminate every delayed delivery, incorrect charge, missed appointment or disputed claim. Recovery deserves to be designed with the same care as the original service.
Five principles matter.
Restore time. Do not make the customer pay repeatedly for the same mistake through chasing, waiting and starting again.
Restore control. Explain what has happened, what will happen next, who owns it and when the customer should expect movement.
Preserve memory. Ask once and remember. A customer should not have to reconstruct the organisation's memory at every hand-off.
Create ownership. The person receiving the problem should own the route to resolution, even when other specialists are needed.
Measure what the customer endured. A closed task is not necessarily a resolved need. Study repetition, chasing, hand-offs, channel switching and avoidable delay.
A leader can begin tomorrow by spending an hour at the front line and listening to ten calls.
Do not score the people. Study the work.
How many customers are chasing an update? How many are repeating information? How many are calling because a promise failed? How much demand exists only because something was not done, or not done correctly, the first time?
Those calls reveal the organisation customers actually experience, rather than the one shown on the dashboard.
What will your customer believe happens next?
My television may arrive today. It may not.
Either way, Amazon has already achieved something valuable: it has made the likely cost of recovery feel low enough that yesterday's failure has not yet become evidence against the whole relationship.
The £8 purchase has done the opposite.
That is the strange economics of forgiveness. A company can lose more trust over £8 than another loses over an expensive television. The monetary value of the failure is not the same as its psychological cost.
Customers forgive organisations when the recovery restores their time, their sense of control and their belief that someone will take responsibility.
They withhold forgiveness when a small mistake becomes a preview of every future interaction.
The most useful question after something goes wrong is therefore not:
“How quickly did we close the complaint?”
It is:
“What did we make the customer believe would happen next?”
If you'd like to read more about customer effort you can get my new book for free at www.loweffort.com/claimser
If you'd like to get your low effort score in 5-7 mins you can get it here at www.loweffort.com (note at present the tool only applies to claims).
Sources
Source · Financial Conduct Authority, Aggregate complaints data: 2025 H2
Source · Financial Ombudsman Service, Half-yearly complaints data: H1 2025
Source · PwC, 2025 Customer Experience Survey
Source · Amazon UK, How to return an Amazon package for free
Source · The Ritz-Carlton Leadership Center, The Art of Service Recovery
Source · Where service recovery meets its paradox, Journal of Service Theory and Practice, 2022
Source · Consumer Intelligence, insurance-retention research reported November 2025
Source · Descartes Consulting, insurer Low Effort transformation results, published in Claims ER and at
