By Mehdi HassanWhat questions should you ask in an on-site survey?
Ask open questions about what stopped the visitor, what they came for and what they're unsure about. "What's stopping you from signing up today?" is the single most useful on-site survey question, because it asks for the objection directly. Keep each poll to one question so busy visitors still answer.
On-site surveys fill the gap that behavior data leaves. Heatmaps and recordings show where visitors hesitate; a poll on the same page tells you why, in their words. That's why we treat surveys as part of the same review routine described in our hub on heatmap analysis. Below is a short question bank, guidance on timing, and the process we use to turn answers into page changes.
What is a website exit survey?
A website exit survey is a small poll that appears when a visitor shows signs of leaving, such as moving the cursor toward the browser bar on desktop, and asks why. Other on-site polls appear after a set time, after scrolling to a point, or after an action like visiting pricing.
Trigger | When it shows | Best for | Watch-out |
|---|---|---|---|
Exit intent | Cursor heads for the tab or back button | Learning objections from people about to leave | Desktop only; mobile uses other signals |
Time on page | After a set number of seconds | Visitors who've read enough to have an opinion | Too early and nobody knows the answer yet |
Scroll depth | After reaching a section | Questions about a specific section, like pricing | Ties the question to content they've seen |
After an action | After a signup, download or booking | Learning why converters said yes | Ask quickly, before they move on |
Always-on feedback tab | When the visitor clicks it | Catching bugs and complaints | Low volume; mostly strong opinions |
Most heatmap and recording tools offer some of these triggers; Hotjar has the most survey options of the common tools, as we compare in Microsoft Clarity vs Hotjar.
Which website exit survey questions work best?
The website exit survey questions that work best are short, open and about the visitor's decision, not your page. Use the bank below, one question per poll, and match the question to the trigger.
Questions for visitors who are leaving
- What's stopping you from signing up today?
- What were you hoping to find on this page?
- Is anything unclear about how this works?
- What would you need to know before booking a demo?
- What else are you considering instead of us?
Questions for specific sections
- On pricing: Which plan would you choose, and what's holding you back?
- On pricing: Is anything about the pricing confusing?
- On a feature page: What problem are you trying to solve?
- On the main landing page: In a few words, what do you think this product does?
Questions for visitors who converted
- What nearly stopped you from signing up?
- What made you decide to try us today?
- Where did you first hear about us?
The question "what do you think this product does?" deserves a special mention. If many answers are wrong or vague, your page has a clarity problem that no button test will fix. Our guide to explaining a complex product on your website covers what to do next.
When and where should you show an on-site survey?
Show an on-site survey on one high-traffic page at a time, at the moment the visitor has enough context to answer. Polls everywhere, all the time, annoy visitors and give you answers that are hard to compare.
Good starting points are exit intent on the pricing page, a time-on-page trigger on your main paid-ads landing page, and a post-signup question for new users. Set the poll to show once per visitor, keep it small and dismissible, and never block the page content with it. On mobile, a small slide-in after scrolling past the hero is less intrusive than a full-screen overlay. Run each poll until answers start repeating themselves, then switch it off and act.
How do you turn survey answers into copy changes?
Turn answers into copy changes by grouping them into themes, counting how often each theme appears, and rewriting the part of the page that each top theme points to. Most themes turn out to be copy problems: a missing answer, an unclear claim or an unstated price.
Export and read every answer
Read them all before categorizing. You'll spot the themes faster than any automated tagging, and you'll notice useful phrasing.
Tag each answer with one theme
Typical themes are price, unclear product, missing feature, trust, timing and wrong audience. Add new themes as they appear.
Count and rank
Sort themes by how often they appear. The top two or three are your fix list.
Map each theme to a page section
Price confusion maps to pricing copy; 'what does it do?' maps to the headline and subheadline; trust maps to proof and security sections.
Rewrite with their words
Use the visitors' own phrasing in the new copy. If they say "I don't know if it works with our CRM", the page needs an integrations answer in those words.
Check the effect
Run the same poll again after the change. If the theme shrinks, the fix worked.
Hypothetical example: from answers to a rewritten pricing section
For example, suppose a SaaS company runs an exit poll on its pricing page asking "What's stopping you from signing up today?".
Reading the answers, the team tags them into themes. The largest theme is uncertainty about what happens at the end of the trial. The next is whether the product connects to tools the visitor already uses. A smaller group says it's too expensive.
The team adds one line under the trial button saying what happens when the trial ends, adds an integrations row to the pricing table, and leaves the price alone for now, because the price theme is the smallest and might shrink once the other doubts are answered.
This example is illustrative. The method is the point: theme, count, map, rewrite in their words, re-poll.
Should you use multiple choice or open questions?
Start with open questions, then switch to multiple choice once you know the common answers. Open answers reveal reasons you hadn't thought of; multiple choice is faster to answer and to count, but only covers the options you already imagined.
A good pattern is to run an open question first, build a multiple choice version from the top themes plus an "Other" option with a text box, and use that version to track how themes change over time. Keep the open version for pages you're about to rewrite, because the exact phrases are what a writer needs. If you're handing answers to a writer, our guides to SaaS website copywriting and hiring a SaaS website copywriter explain how they'll use them.
What do common survey answers usually mean?
Common survey answers usually point to one of a few fixable problems: the page doesn't explain the product, doesn't answer price, doesn't prove it works, or is speaking to the wrong buyer. Each has a typical fix.
- "Not sure what it does": rewrite the headline and subheadline around the outcome, and show the product earlier.
- "Too expensive" or "no pricing": show price or a range, and explain what's included. Hidden pricing often reads as expensive.
- "Need to talk to my team": give them something to share, such as a one-page summary or a short video.
- "Does it work with X?": add an integrations section, using the tool names visitors mention.
- "I'm a developer, where are the docs?": the page serves two audiences; see our guide to selling to technical and non-technical buyers.
- "Just browsing": not a problem to fix. Make sure there's a low-commitment next step, like a newsletter or a demo video.
Industry pages often surface specific doubts. Fintech visitors ask about security and compliance, a pattern our guide to fintech website redesign addresses, while self-serve products hear about setup effort, covered in PLG website design. For the measurement side of any change you make, see how to calculate conversion rate.
