Five User Interviews Beat a Thousand Survey Responses

When a brand wants to know why customers aren't buying, the first instinct is usually a survey. It goes to the whole list, a few hundred people reply, and the results come back as a bar chart. Price is the top reason and delivery cost is second. Nobody learns anything they couldn't have guessed.

At User Actually, we use surveys for counting and conversations for understanding. A handful of well-run interviews, prompted by real session recordings, tells you why people hesitate in a way a multiple-choice question can't.

A survey tells you how many people ticked a box. An interview tells you why they hesitated.

1. Why Surveys Disappoint

Surveys are good at measuring how common something is once you know what to ask. They're poor at discovery. A multiple-choice list can only return answers you thought of in advance, and open text boxes get one-line replies. People also explain their behaviour after the fact, and those explanations tend to be tidier than what happened.

2. Use Clarity Recordings as Your Prompt

Before you speak to anyone, watch what people do. Microsoft Clarity session recordings, filtered to one step such as mobile product page to basket, show where visitors stall, scroll back or leave. Note the moments that puzzle you. Those become your interview questions.

3. Run Five Interviews in a Week

Nielsen Norman Group found that a first usability study with five participants uncovered 85% of the usability problems. Five is enough for a first round, as long as they resemble the customers you care about.

• Recruit recent customers and recent abandoners. Offer a small thank-you, such as a voucher, and say upfront how long the call takes.
• Set a task, not an opinion question. "Find a gift under £50 and get to checkout" shows more than "What do you like?"
• Ask about the last real purchase. Past behaviour is more reliable than predictions about what someone would do.
• Let silences run. People fill them with the detail you need.
• Write down their exact words. Paraphrasing early loses the language you'll later use in copy.

4. Add a One-Question Poll at the Right Moment

Interviews tell you why. A short on-site poll tells you how many. Put a single open question at the step where recordings show people stall, such as "Is anything stopping you from ordering today?" on the basket page. Read every answer, group them by theme and count each group. The poll checks whether what five people told you applies to hundreds.

5. Turn It Into a Journey Map

Pull everything onto one journey map. Across the top, list the stages your customers go through, from arriving on the site to waiting for delivery. For each stage, write what the customer is trying to do, what they're thinking, where they get stuck and the evidence for it, whether that's a quote, a recording or a poll count.

The map should show friction, not decoration. If a stage has no evidence, leave it blank and go and get some.

6. From Map to Priorities

Each problem on the map becomes a candidate fix. Rank the fixes by how many customers the problem affects, using GA4 and the poll, how strong the evidence is and how hard the fix will be. A problem that appears in interviews, recordings and poll answers goes to the top.

Many fixes will involve answering a doubt earlier, which is the job of a good product page. Test the bigger changes rather than shipping them on faith.

The Bigger Point

Research doesn't need a budget line or a six-week project. Watch some recordings, talk to five customers, ask one question on site and put it all on a map. By the end of the week you'll have a ranked list of the problems your customers have, described in their own words.