How to measure customer satisfaction

Transactional: one question within a day of the moment — “how did the delivery go?”, “was the problem fixed?” — with a score and an optional comment, on WhatsApp for most Indian customers, from a person’s number so a reply is possible. Relationship: a short check twice a year, or at the quarterly review for larger accounts — how are we doing overall, would you recommend us, what should we change — asked in a conversation for the top accounts and in a two-question message for the rest. Operational: the numbers you already have — on-time delivery rate, first-contact resolution, complaints per hundred orders, repeat rate, days to pay, and customers who have gone quiet — read monthly alongside the asked signals.

The operational signals are the honest ones: a customer who scores you nine and stops ordering has told you something the score did not.

CSAT — a satisfaction score, usually one to five, on a specific interaction — is the transactional tool: fast, specific, and useful for finding which step or which person is failing. NPS — the recommendation question, zero to ten, with promoters, passives and detractors — is the relationship tool: a rough temperature of the whole base, comparable over time, useful mostly for the follow-up conversation with the detractors and the referral ask to the promoters. Neither number means much on its own in a small business; twenty responses do not make a statistic. What matters is the trend, the outliers, and the comments.

Qualitative feedback — the conversation — is the tool that produces the actions: ten minutes with a customer about what they would change gives more than a hundred scores. For the top accounts it is the quarterly review; for the rest, a call to anyone who scored low and a call to five customers a month chosen at random. Ask, listen, write three lines, thank them.

Closing the loop is the part that changes satisfaction, and the part most measurement programmes skip. Every low score gets a call within a day from someone who can fix things; every comment gets an acknowledgement; every customer who suggested something is told what happened to it. A customer who complains and is called back becomes loyal more often than one who never had a problem; a customer who scores you badly and hears nothing has been told the survey was decoration.

Monthly, the owner or the responsible person reads the pattern: the step or person with low transactional scores, the themes in the comments, the detractors and why, the operational numbers beside them. Three actions, each with a name and a date — a delivery process changed, a person coached, a policy fixed — and next month the same page shows whether the scores moved. Findings without an owned action are a report; findings with one are a management system.

The mistakes: a long survey nobody completes; scores collected and never read; no call to the unhappy customer; chasing the score — staff asking for a ten — rather than the truth; treating a small sample as a statistic; ignoring the operational signals because the asked ones look fine; and no monthly action. A safeguard: for last month’s lowest score, find out who called the customer and when.

A working programme is one question at the right moments, a conversation with the customers who matter and the ones who are unhappy, the operational numbers read beside the scores, every low score answered within a day, and three actions a month with names. This is the feedback programme we set up with SMEs — the moments and questions, the WhatsApp mechanics, the loop-closing rule, the monthly page and the actions — and the free audit starts by asking what happened after the last complaint.