Customer survey questions for B2B: what to ask and which answers to f...

A good B2B customer survey is short. Ask one overall question, one question about why, a few questions about specific parts of working with you, one question about the respondent's role and an optional request for a follow-up call. Ask about recent, specific experiences rather than general impressions. Then call the people whose answers you cannot explain from the survey alone. The survey shows you where to look, and the calls show you what to fix.

A worked example

Suppose you sell appointment scheduling and patient reminder software to about 220 independent dental practices. At each practice, the owner-dentist signed the contract, the office manager runs the account and front-desk staff use the product all day. Contracts renew each year. You want to know which practices are at risk and what to fix first. This example is hypothetical and is used throughout the page.

Decide who at each account should answer

In B2B, "the customer" is several people. Bain's page on Net Promoter in B2B notes that business accounts often include gatekeepers, business leaders and operational users, each with different needs, and that companies have to choose carefully whom to ask for feedback.

For the dental software, that means surveying the office manager and at least one front-desk user at each practice, and hearing from the owner separately, often on a call. Add a role question so you can split the results. A practice where the front desk is frustrated but the owner is happy has a different problem from the reverse, and an average hides both.

Questions to choose from

Pick the rows that matter for your decision. Most customer surveys need fewer than ten questions in total.

What you want to learn	Question	Answer format
Overall satisfaction	"Overall, how satisfied are you with [product]?"	Very satisfied / Somewhat satisfied / Somewhat dissatisfied / Very dissatisfied
The reason behind the rating	"What is the main reason for your answer?"	Open text
Recommendation	"How likely are you to recommend [product] to a colleague?"	0 to 10
Effort	"The last time you contacted our support team, how easy was it to get your issue resolved?"	Very easy to very difficult
Recent problems	"In the past month, did a reminder go to a patient with the wrong appointment time?"	Yes / No / Don't know
Value	"Compared with how you handled scheduling before, how much time does [product] save your practice?"	A lot / A little / No difference / It costs us time
What is missing	"What is one thing you did outside [product] last week that you wish it had handled?"	Open text
Renewal risk	"Is your practice currently looking at other scheduling tools?"	Yes / No / Not sure
Role	"Which best describes your role?"	Owner / Office manager / Front desk / Other
Follow-up	"May we contact you for a 15-minute call about your answers?"	Yes / No

A few notes on these choices:

The recommendation question is the Net Promoter Score question. If you use it, keep the standard 0 to 10 scale so your number means what other people's numbers mean. Net Promoter Score explains the scoring and its limits.
The effort question comes from service research. In Harvard Business Review, Matthew Dixon, Karen Freeman and Nick Toman reported on a study of more than 75,000 people who contacted service centers or used self-service. They argued that how much effort customers had to spend predicted loyalty better than satisfaction or NPS. Treat that as their finding about service interactions, and ask the effort question only about a specific recent contact.
Recent events are easier to report accurately. "In the past month, did X happen?" gets a more reliable answer than "How reliable is our product?"

Word and order the questions carefully

The guidance from Pew Research Center and AAPOR that matters most for customer surveys:

Ask the overall question first. Pew describes how asking a specific question before a general one can change the answer to the general one, and AAPOR recommends general questions before specific ones on the same topic. If customers first answer three questions about support tickets, those questions can shape their overall rating, so it no longer reflects the whole relationship.
Ask about one thing at a time. "How satisfied are you with our software and support?" cannot be interpreted. Split it into two questions.
Avoid agree/disagree statements. Pew notes that some respondents tend to agree with whatever statement they are shown. "Our reminders are reliable: agree or disagree?" invites a yes. Ask what happened instead.
Offer "don't know" and let people skip. AAPOR notes that requiring an answer increases the risk that people quit the survey early.
Keep the wording fixed between rounds. Pew and AAPOR both stress that you cannot measure change if you change the question. If you plan to repeat the survey each quarter, settle the wording before the first send.

Keep it short. NN/g's guidelines for feedback requests say a feedback survey should take around one minute, and recommend telling people how many questions there are. They also suggest putting the optional open-text field at the end, because a blank box at the start can put people off. A quarterly relationship survey can run a little longer, but every extra question lowers the chance that a busy office manager finishes it.

Decide which answers deserve a call

The survey might tell you that a third of front-desk users are dissatisfied. It will not tell you what happened on the morning a reminder went out with the wrong time. The follow-up calls are for that.

Call respondents who agreed to be contacted from these groups:

Dissatisfied users at accounts that renew soon. Their answers carry the most revenue risk.
Mixed accounts. Practices where one role is satisfied and another is not. Ask each person what the other might not see.
Unclear open-text answers. "It's fine but clunky" needs an example before anyone can act on it.
Accounts looking at other tools. Ask what started the search and who is involved in it.
Very satisfied accounts in a segment you did not expect. They may show where the product fits best.

A simple guide for a 15-minute follow-up call:

"You rated [product] as [answer]. Can you walk me through the last time it worked well for you, or didn't?"
"What happened next? Who else was involved?"
"What did you do instead, or how did you work around it?"
"How often does that happen?"
"If we fixed one thing, what would change for your practice?"

Keep the conversation on specific events, not their opinion of your roadmap. Customer interview questions has more wording, and interviewing churned customers covers accounts that have already left.

Read the results with their limits in mind

Your survey hears only from people who choose to answer. In its explanation of three types of NPS, Bain points out that when you survey only your own customers, people who don't like doing business with you may decide it isn't worth their time to respond. In a base of a few hundred accounts, a handful of silent, unhappy accounts can matter more than the average.

Check response rates by role and by account size before reading the percentages. If most answers come from office managers at your largest practices, the result describes them, not your whole base. Label small groups as small: with 30 responses, a shift of several points between quarters can easily be noise.

Your next step

Choose eight questions or fewer from the table, test them with two or three customers to see how they read them, and send. Block time the following week for five to ten follow-up calls.

A customer survey only reaches current customers. If you also need to hear from practices that chose a competitor, or from people in a role you rarely talk to, Instant Expert can find people who match a description you write. You review who it finds, it sends your invitations, and you pay only for calls that get booked. The directory page for operations professionals in dental services is one place to start.