Voice of Customer Research

Scott Cook asked customers who'd just bought software if his team could watch them install it. While building early Intuit products, including Quicken, he wanted to see what happened after the purchase.

A case study in the 2020 book Testing Business Ideas describes this as the start of Intuit's Follow-Me-Home approach. Carry that curiosity into your interviews: ask about something the customer has actually done.

For this example, I've invented bakery café owner Margot, her staff-scheduling tool Quillberry Roster, and the dialogue and records below. She's describing a Thursday evening when she moved a Saturday shift after sharing the rota.

"I kept checking because I didn't know who'd seen the new shift."

She kept picking up her phone to look for replies. Quillberry's team wants "save time" to open its new sales page. Before you turn her sentence into copy, you need to find out what would have let her put the phone down.

Start with a decision the research can inform

Write down the decision your findings will inform. Quillberry's team needs to choose the lead for its sales page. The research question is what made managers seek another way to handle changes to a published rota.

Ask Margot how much time Quillberry saves and you've already steered her toward the answer. Ask about her last shift change, and you leave room for the reason she kept checking. That answer could change which benefit the page leads with.

Keep technical labels such as "perceived adoption risks" in your research plan. With Margot, ask what she checked before deciding to try Quillberry, then follow her answer before introducing a concern from your own list.

This lesson covers interviews and surveys. The broader copywriting research process includes product evidence and alternatives. Review mining can supply public accounts too, though you can't ask a reviewer what they meant.

Recruit people close to the event

Choose participants who lived through the decision you're studying. For Quillberry, screen for managers who recently changed a rota after sharing it. A buyer like Margot can describe what she compared; someone who kept their current method can explain why it was enough.

Long-term customers can help with a different question, such as how the product fits their routine. Record which group each person belongs to, or the notes get hard to read later.

As a practical starting plan, you could invite four recent buyers and two people who declined. Check that each can recall a particular shift-change episode before booking the interview. Note how you found them and whose perspectives are missing.

If the sales team only introduces enthusiastic customers, put that selection limit beside the findings. It explains whose experience the page team is hearing.

GOV.UK's informed-consent guidance asks you to explain what you'll collect and whether you'll record. Say who can access the material and how long you'll keep it. Make participation voluntary, record the agreement, and seek separate permission before publishing anyone's words.

Build a guide around one actual episode

GOV.UK's interview guidance recommends open, neutral questions and accounts of real experiences. It also suggests trying the guide on a colleague first. In your rehearsal, replace "Describe your acknowledgement workflow" with a question a manager can answer about an actual shift.

In an interview you rebuild the sequence from memory, so start with a little context about the person's role. Ask about the last time they changed a rota after sharing it. Find out what happened before you ask what they thought of it.

An interview guide for Quillberry Roster

1. Set the scene

   Tell me about the last time you had to change a rota after sharing it. What prompted the change?

2. Recover the sequence

   Ask these separately: What did you do first? What happened after that? Who else was involved?

3. Explore the decision

   When did you start looking for another approach? What did you try or compare?

4. Check meaning

   Return to an important phrase the participant used and ask what it meant in that situation.

When Margot mentions checking replies at home, follow that before moving to your next planned question. Ask what she did after finding no reply. Her answer will tell you which part of the episode needs a closer look.

Remove the answer from your question

"How frustrating was the old system?" assumes frustration. Ask what the person did and let their description supply the feelings.

"Would you like automatic reminders?" asks for a prediction about an appealing feature. Ask how they handled the last missed reply.

Questions for the rota interview

Questions that leave room: 1

Walk me through the last change you made.

Steering version: How much time did the messy process waste?

Questions that leave room: 2

How did you find out whether people had seen it?

Steering version: Would automatic reminders make your life easier?

Questions that leave room: 3

What did you check before deciding to try the product?

Steering version: Why do you love our simple dashboard?

Each prompt leaves room for an answer the interviewer didn't expect.

If someone says the change went fine, resist digging for the pain you expected. Ask how they handled it, and write down that their current approach worked.

If Margot says she checked twice and then went to bed, keep that scale in the notes. "Spent the night worrying" would enlarge her account. A vivid phrase earns its place by staying faithful to what happened.

Follow the phrase into the task

When Margot says she kept checking, ask "What were you looking for when you picked up the phone?" In our invented interview, she answers: "A reply saying they could work Saturday. Seeing that they'd read it wouldn't tell me whether they could come."

Now you have a finding to explore: she needed agreement to the changed shift. A read receipt alone wouldn't have let her stop checking. Ask what she did while waiting for that agreement before moving to another topic.

Intuit's guidance says to go where the customer experiences the problem and observe first. Then ask open questions, including "Why did you choose to do it that way?" If Margot can show you a shift change, follow that order.

Write down where she looks after publishing it and what she checks next. Wait until she finishes before asking why she opened a separate message thread. As with Cook's installation visits, the task gives you something specific to ask about.

Offer "So you wanted peace of mind?" and you've handed her a tidy abstraction. Her detail gets swapped for your interpretation.

If she calls the process "awkward," ask which step she means. At the end, check your understanding: "You published the change, then opened the chat to look for replies." Ask whether you've got the sequence right so she can correct it.

Keep the sales pitch out of the room. If Margot expects Quillberry to show who has read a change, ask where that expectation came from before explaining the feature. Save product questions for after the research so you can record how she understood the offer.

Use surveys to collect customer language

A short survey reaches people you can't interview. It can't ask a follow-up when someone writes "easier," so the prompt needs a clear event or decision. Give people room to answer in their own words, and let them skip what they can't recall.

Pew Research Center's question-writing guidance shows how wording, answer choices and order can shift responses. After the 2008 presidential election, Pew asked voters which one issue mattered most in deciding how they voted.

When the economy appeared on a list of choices, 58% picked it. When people answered in their own words, 35% named the economy, and 43% of that open-ended group gave an answer the list didn't include.

So for Quillberry's survey, I'd put the open prompt before any list of possible benefits. A list can quietly supply the vocabulary you were hoping to hear.

A short post-decision survey

1. 1Opening: explain the purpose, voluntary participation and how responses will be used.
2. 2What was happening when you began looking for a different way to manage rotas?
3. 3What, if anything, made the decision difficult?
4. 4Which option did you choose? Include keeping the current approach and an open alternative.
5. 5Optional: may we contact you to understand your answer better?

Pew also advises asking about one concept at a time. Ask about ease and affordability separately, since people might answer them differently. Try the form on someone outside the project before you send it widely.

Keep each open question tied to the page decision. "What was happening when you began looking?" can reveal the buying trigger; a general satisfaction rating won't recover that episode.

Counts from a small convenience sample describe the respondents and can't establish market prevalence.

Preserve the words and the conditions around them

When you come back to Margot's phrase in a month, you'll want the question beside it. Store her participant code and decision stage alongside the exact wording. Add the question and a recording timestamp, and keep names in a separate record with restricted access.

Check every excerpt against the recording before you call it verbatim, and mark anything else as a paraphrase. In the record below, Margot is P04.

Evidence record: participant P04

Recent buyer; question: Tell me about the last time you had to change a rota after sharing it.

I kept checking because I didn't know who'd seen the new shift.

Follow-up question: What were you looking for when you picked up the phone?

A reply saying they could work Saturday. Seeing that they'd read it wouldn't tell me whether they could come.

Researcher's interpretation and product check, kept separate

Possible concern: confirming that staff can work a changed shift. Check whether the product shows agreement, a read receipt, or only that a notification was sent. Those support different promises.

Keep the customer's words beside the question that prompted them.

"It saves time" means one thing after an unprompted story and another after "Does it save you time?" You can tag both, and give them different weight. Keep exceptions and negative accounts too, so the most striking phrase doesn't quietly become the only story.

If you use a transcription or AI tool, check the participant's agreement and your organization's data rules before you send any recordings. A generated summary can help you find a passage. Quote from the recording itself, and never stitch a quotation together from several people.

What does the evidence permit?

Margot's follow-up answer points to a need for agreement to work the shift. It can't show that Quillberry meets that need. If the product only sends a notification, "Know everyone has seen every change" would claim more than it does; even a read receipt wouldn't establish agreement.

Compare the visible statuses with the promise the draft page makes.

A product check to bring to the meeting

Change a sample shift with the product team watching. Ask a test recipient to open it without replying. Record what the manager can see, then have the recipient confirm they can work the changed shift and record what changes.

If another manager says sending the notification is enough, keep that account beside Margot's. Ask whether they need a reply agreeing to the shift or simply a record that they sent it. Turning research into copy explains how to choose a message from these different situations.

Prepare a kit for one decision

The Research lessons show where interviews fit in a project. For Quillberry, put the page decision at the top of your guide: which shift-change concern deserves the opening? Under it, keep your episode questions and space for Margot's exact words.

Rehearse the guide with a colleague and underline words that tell the customer how to feel. Replace any question that presumes frustration with a request to describe what happened after sharing the rota.

Write the invitation to Margot: "Could we watch you change the next shift, from publishing it to checking the replies?" Bring her answer about agreeing to Saturday's shift to the product review. Check whether Quillberry can show that agreement before drafting a promise about it.