Quick answer: An onsite survey is a short question, or a few questions, shown to visitors while they are on your site. Used well, it tells you what stopped them buying or getting in touch, in their own words. Ask one open question at a time, show it on the pages where people drop off, time it so it does not interrupt, and read every answer. Tools such as Hotjar polls and Qualaroo make it easy to set up without a developer.
Your analytics can show that most visitors leave the pricing page without clicking anything. It cannot tell you whether the price was too high, the plan names were confusing, or they simply wanted to check with a colleague first. Guessing at the reason leads to changes that may not help. Asking is usually quicker.
Why surveys belong in your conversion work
Most conversion research falls into two groups:
- Quantitative: analytics, funnels, heatmaps and A/B test results. These show what is happening and how often.
- Qualitative: surveys, user testing, customer interviews and support conversations. These show why.
Onsite surveys are one of the cheapest qualitative sources. They reach real visitors at the moment of decision, not people recalling a visit weeks later. That makes them a core part of any CRO audit.
Types of onsite survey
Exit intent surveys
These appear when the visitor moves the mouse towards the close button or address bar on desktop. They are a good fit for “What stopped you today?” type questions. They do not work the same way on mobile, where there is no cursor to track.
Timed or scroll triggered polls
A small box slides in after the visitor has spent a set time on the page or scrolled a certain distance. Because it waits, it reaches people who have had a chance to read.
Page specific polls
Shown only on one page, such as the checkout, pricing or a key landing page. These let you ask very targeted questions.
Post purchase surveys
Shown on the thank you page or sent by email after an order. Buyers can tell you what nearly stopped them, which is often the same thing stopping others.
Questions that get useful answers
The question matters more than the tool. A few that tend to work well:
For visitors about to leave:
- “Is there anything stopping you from buying today?”
- “What information is missing from this page?”
- “What were you hoping to find here?”
For pricing and checkout pages:
- “Do you have any questions about our pricing?”
- “What is your biggest concern about ordering?”
For buyers:
- “What almost stopped you from buying?”
- “What made you choose us over other options?”
- “How would you describe us to a friend?”
Tips for writing questions
- Ask one question first. A single short question gets far more responses than a form. You can follow up with a second question for people who answer.
- Use open questions when you are exploring. Multiple choice answers limit people to the reasons you have already thought of.
- Use multiple choice later to measure how common each reason is, once open answers have shown you the main themes.
- Avoid leading questions. “How easy was checkout?” assumes it was easy. “How was checkout?” does not.
- Keep it conversational. Write as you would speak to a customer in a shop.
Where and when to show surveys
Showing a survey on every page to every visitor annoys people and gives you noisy data. Be selective.
- Find the drop off points. Look at your funnel and goal reports in Google Analytics. The pages where most people leave are where you want answers.
- Pick the moment. Wait until visitors have had time to engage. Asking someone who arrived three seconds ago what is stopping them is pointless.
- Limit frequency. Show the survey once per visitor and stop when you have enough answers.
- Respect mobile users. A survey that covers the whole screen on a phone is a poor experience. Use a small, easy to dismiss box.
Tools for onsite surveys
Several tools make surveys easy to add with a single script:
- Hotjar includes polls alongside heatmaps and recordings, which lets you combine what people do with what they say. We compare it with another popular option in Hotjar vs Crazy Egg.
- Qualaroo is a dedicated survey tool with detailed targeting rules, branching questions and answer analysis.
- Form plugins on WordPress can do simple post purchase surveys if you do not need targeting.
Choose based on how much targeting you need and whether you want surveys alongside other research tools in one place.
How many responses you need
Surveys are not a vote. You are looking for themes, not statistical proof. In practice, after reading somewhere between 50 and 200 open answers, patterns usually start repeating. If the same concern shows up again and again, you have found something worth acting on.
For multiple choice questions where you want to measure proportions, more responses give more reliable numbers, but even then you are looking for clear differences rather than precise percentages.
Making sense of the answers
Raw answers are messy. A simple process helps:
- Export everything to a spreadsheet.
- Read every answer. Do not skim. The exact words people use are valuable.
- Tag each answer with a theme, such as price, shipping cost, trust, missing information, technical problem or “just browsing”.
- Count the themes. The biggest groups are your priorities.
- Note useful phrases. Visitors’ own words often make better headlines and button text than anything a copywriter invents.
Ignore answers like “just looking”. Some visitors were never going to buy, and that is fine.
Turning answers into changes
Common themes and typical responses:
- Price concerns: explain value more clearly, show what is included, compare plans side by side or add a guarantee.
- Shipping and returns questions: show costs and policies early, not just at checkout.
- Trust worries: add contact details, real photos, secure checkout notes and clear policies.
- Missing information: add it. Specifications, sizes, delivery times and FAQs often fill the gap.
- Technical problems: reproduce and fix them. A survey can surface bugs that never show in analytics.
Where possible, test changes rather than simply rolling them out. An A/B test confirms whether the fix actually lifts conversions.
Common mistakes
- Asking too many questions at once.
- Running surveys on every page with no clear goal.
- Collecting answers and never reading them.
- Treating a handful of answers as the final word.
- Forgetting to switch surveys off once you have enough data.
FAQ
Do onsite surveys hurt conversion rates?
A small, well timed poll rarely does. Large pop-ups that block the page can, so keep surveys light and easy to close.
How long should a survey run?
Until you stop seeing new themes. For a busy page that might be a few days. For a quieter site it may take a few weeks.
Should I offer an incentive for answers?
Usually not for a one question poll. Incentives can attract people who answer just to get the reward.
Find out what your visitors are thinking
Surveys are one of the quickest ways to stop guessing about your site. We run onsite research as part of our conversion work, then turn what visitors tell us into changes worth testing. See how we help businesses turn visitors into customers.
