Using exit surveys to inform website decisions

Close up of someone filling out digital survey

The first question I usually get asked at the checkout of my neighbourhood grocery store is ‘Were you able to find everything you were looking for?’. It’s a great question from a customer service perspective. There’s nothing more frustrating than leaving the store without a key ingredient for my famous barbeque sauce simply because it moved from aisle four to aisle nine, and I couldn’t find it. It’s a chance for the store to ensure that my experience is a positive one by locating the tamarind paste for me.

You need to be able to do a similar thing with a transactional website. If your site does anything more than simply serving content pages, chances are there are at least some percentage of your users who are having a difficult time with your interface. If you know what those problems are, you will be better equipped to prioritize them and fix as it makes sense. Being able to ask visitors how to improve on their site experience is one of the best ways to ensure your marketing is focused on the people that it serves.

Setting up funnels and heatmaps

As I noted in a previous post, Testing UX Design Assumptions with Website Heatmaps, the first step to take to improve the overall user experience of your site is to watch how people use it. One of the simplest strategies for doing this is to use a tool like VWO to set up heatmaps and clickmaps to establish a baseline of where people click on any given page of your site. A heatmap shows where your users spend time on your site. Where does their cursor or finger interact with items on the page? The more interactions in a specific area, the more ‘heat’ you see there. Clickmaps show the number of clicks on the links, buttons and form elements within your site. Used together, these tools can really help show how your site is being used.

You can also create funnels within tools like Google Analytics or MixPanel to see where users drop off as they try to get something done on your site. A funnel can be as simple as seeing how many Pay Per Click site visitors take the time to fill out a form for an ebook, or it can be as complex as tracking someone from the moment they add a product to their cart all the way through to the ecommerce checkout receipt page. This will help you to see where the major friction points are in these funnels and allow you to run a conversion rate optimization program to continue to increase your site’s performance.

Sometimes, you need to stop and ask for directions, which is exactly the function of a site exit survey. It can be one of the quickest ways to confirm an assumption that requires significantly less traffic than an A/B test. Split testing, while still one of the very best ways to determine which improvements to make based on actual completion data, rarely answer the question ‘Why?’ definitively. Plus, sometimes you want to hear the qualitative answer from actual site users as to why they’re not completing the funnel you have set out for them. Exit surveys are also an excellent way to gain greater insight into broader issues with your web presence, or to find out what features visitors want to see in your site.

Anatomy of an exit survey

At Kula, we use Visual Website Optimizer (VWO) to setup on-page surveys, but other tools can also be used such as Survey Monkey or even a basic Google Form. Once the survey has been triggered, it will show up in the location that you specify within your page layout, often in the lower right corner or somewhere similar. Most tools allow you to customize the design of the survey to match your site as closely as possible.

Try to keep your survey as simple as possible, and only show it to users who are at a specific stage within your funnel, such as just before the checkout page of an ecommerce site. You don’t want to annoy 100% of your site visitors with a survey request that may not even apply to their needs. A tool like VWO makes it possible to only show the survey to those users who meet the criteria you set out. You can choose, for example, to only show the survey to specific traffic groups such as those people who have arrived at your site via a certain channel such as email or social networks.

The next step is to determine the triggering event. It could be the clicking of a particular button on the page, it could be the time spent on the site, or the fact that they scrolled all the way to the bottom of the page.

Once you have your target traffic group configured, it’s time to formulate your survey questions. If you’re looking to survey a large group on a more general site page, you might keep the question and answers fairly simple, although it’s unlikely that you’ll see much response to open-ended, freeform questions without a defined answer set. Questions like ‘How can we improve this site?’ won’t help you fix specific problems in your funnel, but they could provide you with some context around what visitors might be looking for. Keep the survey as simple as possible – a single question if you can – otherwise you can anticipate a lower response rate.

Writing a good exit survey question

While questions as general as the one noted above can be too simple for you to action (since they have no focus, the answers will be all over the place), something as binary as a yes or no question can be problematic as well. Our best results with exit surveys have come from the use of a single question with multiple answers specified and an ‘other’ field for freeform answers. VWO will parse the results from free text fields and show you the most commonly used words so that you can see at a glance if the answers align with your expectations or contain trends.

The best questions tend to centre around how your site can better serve your visitors. How can you do better. When people know you’re listening, they’re more likely to feel positive about the experience, even if they’re unable to find what they’re looking for. One survey we conducted simply asked ‘What stopped you from submitting a quote request today?’

The answers were as follows:

  1. I prefer to purchase this product in person
  2. There was no pricing information available
  3. I need more information
  4. I can’t buy the product I want online
  5. Other

The overwhelming response, at 66.67%, was #2. Due to supplier requirements or other issues, not all retailers were able to display pricing information online. Although we suspected that this was the biggest conversion blocker, we now knew with relative certainty that this was the case. Armed with this information, our client has been able to make larger changes to their online presence that should increase conversions and better serve their visitors.


This kind of broad qualitative result would be very difficult to get via heatmaps or even A/B testing. While it is possible, it would require a much larger sample size and a number of successive split tests. Even in-person UX testing might not uncover this particular issue since it only presents itself for those users currently shopping for this particular product. It’s the perfect kind of exit survey, and can make a serious positive impact for your leads and customers if actioned appropriately.

What if you can’t action the survey results right away?

Simply running an exit survey is a great start, but if you’re unable to implement the changes the results suggest, what’s the point? It might provide you with the business case necessary to pitch your boss, suppliers or others within the organization on a new technology platform or a change in business strategy. It may help you to determine the options you want to test in an upcoming A/B test. Or, it may simply give you food for thought to consider for future iterations of your site.

In addition, there’s a chance that your results could be inconclusive. You might not have enough responses with which to reach a consensus or there could be a tie between several of the results. In this case, refine your questions and/or answers and run the survey again until you achieve statistical significance.

Next steps on the road to improving your conversion rate

If you are able to quickly follow a positive survey completion with an A/B test to confirm your hypothesis from the results, so much the better. Exit surveys are an excellent way to quickly test your assumptions, see if the answers are trending in the direction you anticipated or not, and focus on what to change from there. They aren’t a substitute for split tests, or other forms of qualitative testing, but they are excellent for establishing a direction with which to start improving your site.

Join us in conversation…