NPS in a small business: is a simple score worth collecting

A consultant tells you to start measuring NPS, and you wonder whether a score out of ten is worth the trouble in a firm of twelve people. It is worth collecting only if you will act on the low answers. If you will phone everyone who scores you six or below, the question pays for itself in the first round. If you will paste the number into a review and never speak to a detractor, skip it. The phone call is where the value is, and the score is just the way you choose whom to ring.

What does the question ask?

One question: "How likely are you to recommend us to a friend or colleague?", answered from zero to ten. Nine and ten are promoters, seven and eight passives, zero to six detractors. Promoters minus detractors, as percentages, is the score. It runs from minus a hundred to plus a hundred. A second question, "What is the main reason for your number?", is where most of the value sits.

Should you ask at all?

Decide on facts about your customers.

Ask if you have at least twenty or thirty repeat customers and someone has time to phone five or six after each round. Know every customer personally? Then do not. Ring them and ask directly. If customers buy once and never return, ask a closing question at hand-over instead. If customers keep mentioning a problem you have not fixed, fix it first, since the score would only record it. Business customers with several people involved may need two askings, one to the user and one to the buyer, because their answers can differ.

How do you send it?

Keep it to two questions, from a named person and not "the team", in the channel the customer already uses with you.

Hello Prakash, Arvind here from the hardware counter. One quick question, no forms: on a scale of 0 to 10, how likely are you to recommend us to another contractor? And what is the main reason for your number? Just reply here.

A link to an outside survey page often gets fewer replies than a plain answer to a message. Ask after a normal stretch of service, not straight after a problem, or you measure the problem. And say who reads the answers only if it is true. "I read every reply myself" has to be so.

How do you read thirty answers?

Ignore the percentage first. Make a list: name, score, reason. Then read in an order.

Detractors come first. For each, write one line: what went wrong, whose job it was, whether it can be fixed. Passives next; their reasons often sound like "fine, but no reason to switch either", which tells you about differentiation. Promoters last, for their words. A promoter's phrase may suit your next proposal better than your own wording, if they agree to be quoted.

Then count the reasons, not the scores. If five of eight detractors mention delivery, you do not need the score to know where next week's effort goes.

What do you say to a detractor?

Call within a few days of the answer, and make it the owner or a senior person. Thank the customer, ask for more detail, and offer to deal with whatever is named. Do not argue or explain. For example: "You gave us a four and said deliveries were irregular. I would rather hear it from you than guess. What happened on the last two orders?" Do not treat the number as an accusation. The customer has opened a door.

A customer who scored you three and heard "that was our mistake, I would like to fix it" may end up a firmer supporter than someone who never had a problem.

The counter that asked by email

Here is an invented case. A hardware and sanitaryware dealer in Siliguri supplies thirty small builders. The owner emailed all of them a survey link. Four replied, three with nines, one with a two and no reason. The headline score looked wonderful and told the owner nothing.

The owner tried again the next quarter in the way above: a WhatsApp message in the name of the counter manager, who knew each builder, with the two questions. Twenty-two replied. Fourteen were promoters, five passives, three detractors. Two of the three detractors named the same thing: pipes loaded on the lorry in the wrong lengths, found only at the site. The owner rang both, fixed the loading check at the godown, and learned that one passive only bought from the dealer because the shop sat on the way to the site. One of the two detractors still moved some orders elsewhere.

The change was at the godown, not in the score. The first attempt failed because of how it was sent: the second reached people through someone they already spoke to.

What goes wrong?

Asking too often. Customers asked every month tend to stop answering; ask after a stretch of service and not again until you have acted on the last round. Paying staff on the number. They may steer customers towards nine and ten, so reward the follow-up calls instead. And treating the number as the goal. The goal is fewer customers leaving without telling you why. A single high score in a small sample can also hide a customer about to leave, so read the comment before the digit.

Count your repeat customers

Count them and ask whether anyone can ring a handful after each round. If yes, send the two questions to your twenty most recent buyers this week. If not, speak to five customers instead, as in how to measure customer satisfaction. If you are unsure whether a score suits your business, tell us what you need; where feedback fits in your wider selling is covered under sales.