Product survey questions: ask about what just happened, then call a few people
Good product survey questions ask about a moment that just happened, from setup to cancellation. Ask three to five, keep it to a minute, then call a few respondents.
Product survey questions work best when they ask about something that just happened: setup, a feature someone used a minute ago, a cancellation. Ask three to five of them, keep the survey to about a minute, and invite a few respondents to a follow-up call. The survey tells you how common an answer is. The calls tell you why.
This page covers people who already use your product. For a market you don't serve yet, see product research survey. To check whether the product has found its market, see the product-market fit survey.
The example: an app for chasing client paperwork
Here is a made-up example used throughout. You run an app small accounting firms use to collect client documents at tax time. About 400 firms use it. You want to know why many firms stop sending requests after their first month, and whether to build the e-signature add-on a few have asked for.
1. Tie every question to a moment
After setup
- "What were you hoping [product] would help your firm do?" Open text
- "Have you sent your first document request to a client yet?" Yes / No / Not yet, but I plan to
- If no: "What has stopped you so far?" A short list built from support tickets, plus "Other (please specify)"
- "Before [product], how did your firm collect documents from clients?" A short list, plus "Other"
Right after someone uses a feature
NN/g's guidelines for feedback requests sum up the timing as "task, then ask": wait until someone has finished a real task. (Nobody wants a pop-up halfway through an upload.)
- "Did the client reminder do what you needed just now?" Yes / Partly / No
- Optional: "What were you trying to do?" Open text, with a character limit
A regular check-in, every quarter or so
- "In the past week, did you use [product] to request documents from a client?" Yes / No. One yes/no question per task beats a "select all that apply" list: Pew Research Center found forced-choice questions tend to give more accurate answers and generally avoids select-all lists.
- "What is one thing your firm still does outside [product] during tax season?" Open text
- "How would you feel if you could no longer use [product]?" Very disappointed / Somewhat disappointed / Not disappointed. This is the product-market fit question; read the linked guide before interpreting it.
At cancellation or downgrade
- "What is the main reason your firm is canceling?" A list from past cancellations, plus "Other"
- "What will your firm use instead?" Open text
- "May we call you for 15 minutes about this?" Yes / No
Before building something new
Describe the feature in two or three plain sentences without selling it, then connect it to what people already do:
- "The last time a client needed to sign something you sent, how did you handle it?" Open text
- "How well would this fit the way your firm works today?" Very well / Somewhat / Not very well / Not at all
- "Who at your firm would decide whether to add this?" Owner / Partner / Office manager / Other
Leave out "Would you use this?" and "Would you pay for this?" A yes costs nothing, so you will collect plenty. Validating feature requests and researching willingness to pay cover tests that ask for a real action.
2. Keep it to about a minute
NN/g recommends about one minute; the two to three minutes many companies ask for is too long. Say how many questions there are, put the optional open-text field last, and ask in the channel where the interaction happened instead of defaulting to email.
Two wording habits from Pew:
- Randomize lists with no natural order. People completing a survey on their own tend to pick options near the top.
- Keep wording identical between rounds. To compare this quarter with last quarter, change nothing in the question or its position.
3. Treat the survey as a count and let the calls explain it
NN/g calls surveys the one common research method that is both quantitative and attitudinal: good at measuring how many people hold a view, poor at explaining behavior. Its example: analytics can show that people abandon carts, but "why" needs qualitative methods such as usability testing.
In the accounting example, suppose 40% of firms that stopped sending requests chose "clients didn't respond." That is a count. Whether clients ignored the message, couldn't find the upload link or preferred to hand over a shoebox of receipts is a question for a conversation.
4. Pick the follow-up calls on purpose
From the people who agreed to a call, choose:
- Firms that stopped after the first month, split by the reason they chose
- Firms whose open-text answers you couldn't interpret
- Firms that asked for e-signature, plus a few that didn't
- Heavy users in a type of firm you did not expect
A 20-minute call can follow this outline:
- Context (3 minutes). "Tell me about your firm and who handles client documents at tax time."
- The event (7 minutes). "In the survey you said [answer]. Walk me through the last time that happened." Keep asking what happened next.
- The workaround and its cost (4 minutes). "What did you do instead? How long did that take?"
- The concept, if relevant (4 minutes). Describe it in the survey's words. "Think back to the last time a client had to sign something. Where would this have fit?"
- Close (2 minutes). "Is there anything I should have asked? Is there someone else at your firm I should talk to?"
Ask before recording (recording consent covers how). Write notes in the same format after every call. Analyzing customer interviews explains how to turn a stack of notes into a decision.
Your next step
Pick the moment that matters most for your current decision, write three to five questions for it, and plan five to eight follow-up calls before you send the survey.
A survey only reaches current users. To hear from people who built or ran similar products, 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 product professionals in accounting software shows the kind of people you can reach.