A customer can tell a retailer something its internal reporting has missed. The account may reveal a helpful intervention, an avoidable frustration or a pattern that deserves attention. I want those accounts heard. I also want the company to be careful about the moment when a description of a shopping experience becomes a judgement about an individual employee. That transition requires more than a calculation.
The customer experiences a whole system through a particular encounter. Product availability, price, returns policy, store layout and the actions of several colleagues may all be present. The person whose name appears in the feedback may have influenced only part of the result. A manager who treats the whole account as an individual performance measure can assign responsibility to the most visible person rather than the person or process able to change what happened.
The whole experience is not one person’s decision
This is not an argument for dismissing criticism. A relevant account can be serious even when it is incomplete. It may require prompt attention through the appropriate process. The point is that management should understand what conclusion the evidence can support. An employee may have acted poorly within a sensible policy, or acted responsibly within a policy the customer dislikes. Those situations need different responses. A single numerical conversion can conceal that difference while making the outcome look consistent.
In 2023 I supported a fictional pilot across thirty-six Thalverin stores that used customer feedback as an automatic input to individual performance ratings. The intention was to give service behaviour more weight alongside sales. Eighty-three ratings were contested. When we examined those cases, the accounts varied substantially. Some concerned direct employee choices. Others involved unavailable products, decisions made elsewhere or interactions that could not fairly be attributed to one colleague without further examination.
We revised twenty-nine ratings and retained fifty-four after reviewing the available material. Those figures matter because they resist two easy stories. It would be wrong to say that every challenge proved the original rating mistaken. It would also be wrong to say that the existence of some well-supported ratings justified the automatic process. The review found individual decisions requiring correction and a broader design requiring change. Management needed to respond to both.
Consistency of calculation can conceal inconsistent judgement
I had mistaken consistency of calculation for consistency of judgement. A formula applies the same operation to each input, but the inputs may describe very different kinds of event. One account may concern a personal choice; another may express dissatisfaction with company policy. If the process treats both as equivalent evidence of individual performance, arithmetic has standardised the appearance of the decision without establishing that the decision itself is sound.
Operations colleagues worried that withdrawing the automatic input would weaken attention to customers. That was a legitimate objection to address. We retained relevant customer accounts as evidence and required an accountable manager to explain the employment conclusion alongside other material. The change did not make customer experience optional. It made the use of that experience more explicit. A company should be able to care about the customer and still examine which part of a difficult encounter belonged to the employee’s responsibility.
The redesign cost USD 160,000, including preparation and review support. There was also the less easily summarised work of explaining why the process had changed. Employees who had challenged a rating needed a response through the proper channel. Managers needed examples of the distinctions they were now expected to make. We could not fix the earlier message merely by replacing a field in the system and assuming that everyone would infer the new standard from the interface.
An opportunity for the employee to give an account is part of that standard. It is not a promise that the employee’s interpretation will be accepted. It allows the decision-maker to consider relevant facts, including circumstances the customer may not have seen. The process must also respect privacy and use qualified advice where required. A public leadership story should describe the management choice without publishing identifiable complaint records or turning an employee’s personal circumstances into evidence of the leader’s sensitivity.
The manager’s task is not always to reconstruct every detail with certainty. Some situations remain ambiguous. The responsible account should say what is known, what is disputed and why the chosen response is proportionate to the evidence available. Pretending that a score resolves ambiguity can encourage a stronger conclusion than the material supports. It can also make the next review less useful, because the uncertainty has disappeared from the record rather than been considered.
Decide what the measure is for before launching it
I now ask how a proposed measure will be used before asking how neatly it can be displayed. Will it prompt a conversation, determine a payment, influence an appointment or become part of a formal performance record? Those uses carry different consequences. A retailer may collect the same customer account for several purposes, but it should not assume that the evidence is equally sufficient for all of them. The intended decision should shape the review, rather than be added after the measure has become familiar.
There is a practical test before launch: examine the awkward cases. What if a colleague correctly applies an unpopular policy? What if the account praises one person for work performed by several? What if the customer’s description is specific but cannot be reconciled with the available record? These questions do not need to produce a universal answer. They should reveal whether the proposed process has a responsible person and a credible route for deciding what to do next.
The result of our pilot correction was a different process and revised records where the evidence justified them. It was not proof that customer satisfaction improved or that employees trusted every subsequent decision. Those claims would need their own evidence. The principle I retain is narrower: customer feedback deserves attention, and an employment judgement deserves an accountable explanation. Treating the first as an automatic substitute for the second can weaken both the customer’s contribution and the fairness of the management decision.