The best customer survey questions for finding test ideas are open-ended questions about the moment someone nearly didn't act: "What almost stopped you from buying today?", "What's the one thing you were unsure about?" and, on an exit survey, "What's missing from this page?"
Ask them at the right moment (just after purchase, or just as someone leaves), then turn the most repeated answers into on-page copy and A/B test it.
That last step matters. A survey only tells you what people say.
A test tells you whether answering it changes what they do.
The question lists below come with real tests from our database where survey answers became winning (and sometimes flat) variants.
Why surveys are one of the cheapest sources of test ideas
Most weak test ideas come from guessing what visitors care about.
Surveys replace the guess with your customers' own words, which you can often paste into a headline, a bullet or a review block almost unchanged.
Oliver Kenyon's team added three reassurance lines above a collection page grid for a store selling wild bird food.
All three lines came from post-purchase surveys, and the test was credited with $44,890 a month ($0.66 per session).
Kasper Dolk's team put survey objections ("is it durable?", "is it the official one?") into benefit bullets in a product page hero, worth a reported +$763,679.14 profit per year.
Surveys sit alongside other qualitative methods. If you need the why behind a single answer, a customer interview goes deeper.
The customer research hub covers the other methods and when each one fits.
Exit survey questions: asking visitors why they're leaving
An exit survey appears when a visitor looks like they're about to leave.
It catches people who didn't convert, which post-purchase surveys can never reach.
How "about to leave" is detected depends on the device.
Contentsquare (which now owns Hotjar) explains that desktop exit-intent surveys usually trigger on mouse movement toward closing the tab, while phones have no cursor, so tools rely on signals like scrolling up, pressing back or a period of idle time.
The same guide notes that Contentsquare's own exit-intent surveys only work on desktop.
On mobile-heavy stores, check what your tool actually does before you trust the sample.
Our exit-intent guide covers the trigger in more detail.
Keep exit surveys to one question, with an optional follow-up. Good ones:
- "What's stopping you from buying today?" (open text)
- "Did you find what you were looking for?" (yes/no, then "What were you looking for?" if no)
- "What's missing from this page?"
- "What would you need to know before you'd sign up?" (SaaS and lead gen)
- "Are you comparing us with anyone else? Who?"
Avoid "Why are you leaving?" on its own.
Many leavers are just browsing, and the question reads as an accusation.
Asking what's missing or what's stopping them gets answers you can act on.
Website survey questions by page and goal
A website survey (also called an on-page or on-site survey) runs while people are still browsing.
Tie each one to a page and a decision you're trying to understand, rather than running one generic survey everywhere.
| Where | Question | What it finds |
|---|---|---|
| Homepage | "What brought you here today?" | Intent mix; whether your hero matches why people arrive |
| Collection or category page | "What's the most important thing when choosing a [product]?" | Filter, sorting and product card priorities |
| Product page | "Is there anything you'd like to know about this product that you can't find?" | Missing specs, sizing, shipping and trust information |
| Pricing page | "What's unclear about our pricing?" | Plan confusion and price objections |
| Cart or checkout | "Is anything making you hesitate?" | Shipping costs, payment options, security worries |
| Signup or form page | "What would make this easier to fill in?" | Field friction and privacy concerns |
| In-app (SaaS) | "How would you feel if you could no longer use [product]?" | Which users depend on you, and why |
The last row is the product-market fit question popularized by Sean Ellis.
In a Venture Hacks interview, Ellis said that if more than 40% of users say they'd be "very disappointed," Ellis tends to call it product-market fit.
For CRO, the useful part is the follow-up: ask the "very disappointed" group what the main benefit is, and use their words on your landing pages.
If you want a loyalty score instead, Net Promoter Score asks how likely someone is to recommend you on a 0-10 scale.
Fred Reichheld introduced it in Harvard Business Review in 2003.
A replication by MeasuringU found NPS explained about 38% of the variation in two-year company growth across seven industries: predictive, but less than the original title suggests.
A score alone also gives you nothing to test.
The open follow-up ("What's the main reason for your score?") is where the ideas are.
Post-purchase survey questions that turn into tests
Post-purchase surveys run on the thank-you page or in a follow-up email.
Buyers have just made the decision, so they still remember what nearly stopped them.
Our post-purchase surveys element page covers placement and tests; this section is the question list.
- "What almost stopped you from buying today?" The single most useful question for test ideas. Answers map directly to objections you can address on the page.
- "What convinced you to buy from us rather than somewhere else?" Gives you your real differentiators, in customer language, for headlines and USP bars.
- "Did you have any questions you couldn't find answers to?" Finds gaps in product pages and FAQs.
- "How did you first hear about us?" Mostly for attribution, but it also tells you which traffic sources to segment tests by.
- "Which other options did you consider?" Shows who you're compared against, and what to say about it.
- "What's the one word you'd use to describe [product]?" Cheap headline and ad copy material.
Here is what that looks like in practice.
Kasper Dolk's team surveyed a supplement brand's customers and kept seeing "what if it doesn't work", "is it a scam" and "the price is high".
They added a "how you'll feel" timeline section to the product page, with risk reversal at the bottom, and reported CVR +14% (still being confirmed when they posted it).
In another case, a post-purchase survey for a Dutch audience said the checkout countdown timer felt like too much pressure.
Kasper Dolk's team tested removing the timer: CVR +25.61%, revenue per visitor +26.18%, an estimated +€150k a month.
That's a survey overturning a common best practice, which is exactly when it's most valuable.
Oliver Kenyon's team pulled objections from post-purchase surveys into an objection-handling review block below the mobile CTA: +$63,735 per month and +$0.43 revenue per session.
@oliverkenyon@oliverkenyon
@kasper_dolk@kasper_dolk@kasper_dolkSurvey-sourced ideas don't always win. A more visible free-shipping threshold from Clean Commit, tagged with post-purchase surveys in our database, came in at +2.4% conversion but inconclusive.
And Clean Commit's shipping info banner, built to answer shipping objections, showed +1.5% conversion but -7.8% revenue per visitor.
An objection people mention isn't always one that changes their behavior, which is why you test.
How to write survey questions that don't mislead you
Small wording choices change answers a lot. Pew Research Center's guide to writing survey questions gives a clear example: when voters were asked which issue mattered most in the 2008 election, 58% chose the economy when it was offered as an option, but only 35% named it unprompted in an open-ended version.
- Start open. In the same Pew example, 43% of people in the open version gave answers that weren't on the closed list. For test ideas you want those unexpected answers, so use open text first, and only build multiple-choice options once you know what people actually say.
- Avoid agree/disagree statements. Pew notes that people tend to agree with statements put to them (acquiescence bias). "This page answered my questions: agree or disagree?" overstates satisfaction. Ask "What questions did this page leave unanswered?" instead.
- Don't lead. "How much did you love our fast shipping?" assumes the answer. "How did delivery go?" doesn't.
- One idea per question. "Was the price and the shipping clear?" can't be answered honestly if one was clear and the other wasn't.
- Watch question order. Pew also documents order effects, where an earlier question shifts answers to a later one. Put your most important open question first.
- Keep it short. One to three questions on-site; five or six at most in a post-purchase email.
From survey answers to A/B tests
A pile of answers isn't a test plan yet. A simple process:
- Collect enough answers. There is no fixed number. Keep collecting until new answers stop adding new themes; for a small store that may take weeks.
- Tag every answer. Group by theme (price, trust, shipping, sizing, "does it work"). Count how often each appears.
- Write a hypothesis per top theme. For example: "Because 1 in 5 buyers said they worried the product was fake, adding 'official licensed product' to the hero will raise conversion." Our hypothesis guide has the format.
- Use the customers' words. The bird food test worked with plain lines like "Packed the week you order it." Don't polish the language until it sounds like marketing again.
- Put the answer where the doubt happens. Shipping worries belong near the price and add-to-cart button, not on an FAQ page nobody opens.
- Prioritize and test. Rank themes by frequency and by how much traffic sees that page. See test prioritization for scoring methods.
To run the surveys themselves, compare options in our website survey tools roundup.
Then browse all tests in the database for what other teams did with similar objections.