Ask the customers your inbox never shows you

A customer follow-up becomes more useful when you can describe who had a chance to answer, what they were asked and whose experience remains unknown. Begin with a service decision and a defined group of people. Keep their replies connected to that group.

Chapter 16 of Infinite Upside follows customer feedback through to a change in the operation and a check on what happens afterward. Gathering more information can help choose that change.

Choose the experience you want to understand

Imagine an independent bicycle workshop reviewing four weeks of completed service. It wants to know whether customers understood how additional work would be approved. This workshop and every count below are invented.

Its list contains sixty collections by sixty different customers. The unit is one customer's completed service visit. The list includes forty weekday collections and twenty Saturday collections. It does not cover people who could not get an appointment, declined a quote or canceled before service. Those experiences could deserve another inquiry; they are outside this one.

The manager keeps the list rather than starting with whoever is easiest to remember. Each row identifies the visit, the contact route available for this follow-up and the eventual outcome of the invitation. A customer returning several times would need an explicit selection rule before the list could still be described as sixty different people.

AAPOR's survey-planning guidance begins with the research question, the people it concerns and the information needed to reach them. Here, the completed-service record supplies a starting list. It does not supply sixty opinions.

Keep the contact record beside the answers

The workshop offers a short, voluntary follow-up through an existing customer channel where that contact is appropriate. Five customers have no suitable route for it. They stay on the original list, marked as not invited. They do not disappear from the account of whom the workshop served.

At the announced closing date, the final states are:

Each person occupies one row category at the close. Someone who began and later finished belongs among the twenty-two completed responses, not in both response categories. Fifty-five invitations were sent, five failed delivery and twenty-seven people returned at least one answer. A delivery report does not establish that somebody read the invitation.

Keep a declined invitation distinct from a delivery failure or an unanswered message. Respect the customer's choice. The record helps explain the evidence; it is not a list of people to pursue until they agree.

AAPOR's Standard Definitions explains why recording how cases ended matters and why a response rate alone cannot establish the amount of nonresponse error. These simple counts are a transparent description of this follow-up, not a claim that its respondents represent every customer.

Ask about one visit in words a customer can answer

“How much better was our new process?” assumes an improvement and asks the customer to endorse it. The invitation shows the collection date and service reference for the selected visit. Each question refers to that service. The workshop could begin with an open question, then ask about the approval step.

• Thinking about that service visit, what, if anything, would you change?
• Was it clear how you could approve or decline additional work? Offer “Clear,” “Not clear,” “Don't remember” and “No additional work was discussed,” as well as the ability to skip it.
• If a step was unclear, what happened?

These are original illustrative questions, not a validated questionnaire. Ask a few people like those who will receive it what they think each question means before using it more widely. If a customer interprets “additional work” as the original service, the answers will not describe the distinction the manager intended.

Pew Research Center's question-writing guidance emphasizes clear, specific wording and asking about one concept at a time. A question about whether a repair was fast, inexpensive and well explained would leave an unfavorable answer difficult to interpret.

Read the answers against the list you began with

Of the twenty-seven people who return something, twenty-three answer the approval question. Six say that the step was unclear. The other seventeen choose another response option; they should not all be described as saying it was clear. Four people who answered something leave this question blank.

The supported statement is that six of the twenty-three people answering this question reported an unclear step. It is not that only six of sixty customers had a problem. The workshop has no answer to that question from the other thirty-seven.

The visit records reveal another limit. Twenty-five of the twenty-seven respondents collected on weekdays; only two collected on Saturday. The original list contained forty weekday customers and twenty Saturday customers. The manager has heard much less of the Saturday experience.

That does not show that Saturday customers were less satisfied or explain why they did not answer. It gives the manager a reason to examine what the follow-up has missed before describing all sixty visits. In sections 8 and 9 of its 2016 survey-quality report, AAPOR discusses comparing respondents and nonrespondents using information already available about them. Knowing that the two groups differ does not, by itself, measure the missing opinions.

Give the findings a next job

The six accounts can lead to useful work now. What did the workshop tell those customers? Did the estimate and later message describe the same approval process? Were they confused about whether work had already started, what it would cost or how to say no? Read the accounts before assigning all six to one cause.

A customer still waiting for help needs that help separately from the research. The complaint-repair example follows the next step: choose a change that addresses the failure actually found, then see whether it works in ordinary service.

If the workshop follows another period, retain the question wording, contact method and definitions needed for a useful comparison, and record any change to them. More favorable replies could accompany a different customer mix, staffing or response pattern. A before-and-after count alone does not establish what caused it.

For your own follow-up, put the completed-service list, the question and the decision it could change on the same page. Afterward, show the contact outcomes alongside the findings. The next person should be able to tell what the business learned and which customers it still has not heard from.