A small popup that uses a single question to ask visitors who are about to leave your site why they are abandoning their purchase lets you learn, straight from the source, the reason for leaving that your statistics do not show. We recommend opening the exit survey on cart and product pages first, at the moment the visitor is leaving; for your first setup, you can make do with one question, a few ready-made options and a short “Other” field. Evaluate the answers separately by page type and device, leave the “I was just browsing” answers out of the count, and address the reason that comes out on top with a single change to your site.
During setup, you decide three things: which pages the survey opens on, when and how often the popup appears, and how many questions and options it has. Once answers have built up, you decide which reason for leaving to tackle first and whether the change you made to the site worked. Four weeks of cart page answers from a hypothetical home textiles store illustrate that second stage with numbers.
Ask the visitor why they are leaving
Your visitor statistics show how many of the people who reach the cart page do not go on to checkout, but they cannot show you why. The shipping fee may have come as a surprise, the delivery time may have seemed too long, or the visitor may simply have been browsing; in all three cases the figures look identical. A well-built e-commerce exit survey, on the other hand, puts that question directly to the visitor who is about to leave. We compared which question the funnel report, the heatmap and the survey each answer in our article on reading customer behavior in e-commerce.
Pushouse’s exit-intent popup module shows visitors a short survey at the moment they are leaving. In it, you can ask them whether they found the information they were looking for and whether price, delivery or something else influenced their decision.
You can draw on general research when you draft the survey options. According to the quantitative study by the Baymard Institute on why carts are abandoned, 42% of US online shoppers give “I was just browsing / not ready to buy” as the reason they abandoned a cart. The institute then sets that group aside and ranks the remaining reasons separately: the most frequent, at 40%, is that extra costs such as shipping and taxes were too high, followed by slow delivery (20%), not trusting the site enough to hand over card details (19%) and being made to create an account (18%). The list has eleven reasons, and the percentages add up to more than 100. Do not compare these percentages with the shares in your own survey; use the list instead as a starting point for your options.
Place the survey on cart and product pages first
Asking the same question on every page lumps visitors who have little in common into a single pool of answers. A visitor who leaves from the home page and one who leaves after filling a cart have stalled at different points. Open the survey first in the two places where purchase intent is highest, and collect answers separately for each page type.
- On the cart page, the visitor has chosen a product and seen the cart total. Expect most answers here to concern the shipping fee, delivery time, payment options and the sign-up step. Some obstacles, such as a shipping fee that stays hidden until checkout, can be removed with a small change to the page; we suggest starting your fixes here.
- On the product page, the visitor is still making up their mind. If the size chart, product dimensions, material, images or stock information are missing, you will learn about it most clearly here.
- On category and search results pages, the real question is whether visitors found what they were looking for. Add these pages after enough answers have built up on the cart and product pages.
- Do not open a survey inside the payment form. A popup that appears in front of someone typing in their card details can be mistaken for an error message and may cause a nearly completed order to be abandoned.
Leave surveys on the home page and blog posts for later. Visitors who leave from there may not have looked at a product yet; expect the answers to cluster on the “I was just browsing” option. Below are two sample popup texts, one for the cart page and one for the product page. Both are written for desktop; change the options to match your own shipping, payment and product terms.
What stopped you from completing your order?
The shipping fee is higher than I expected · The delivery date is too late · The payment option I want isn’t available · I was just browsing · Other (tell us in a few words)
What stopped you from adding this product to your cart?
Not enough size or measurement information · I couldn’t see the delivery time · I’m comparing it with other sites · I was just browsing · Other (tell us in a few words)
On phones, remove one option from each example and keep four: you can drop “The payment option I want isn’t available” from the cart survey and “I’m comparing it with other sites” from the product survey. Keep track of the topic you removed through what visitors type in the “Other” box. If you want to see the price of the product itself separately from the shipping fee, also add a “The product price is too high” option to the desktop cart survey; that brings the number of options to six.
Open the popup as the visitor leaves, and do not show it to the same person too often
What makes a survey answer valuable is that the question is asked close to the moment the visitor makes their decision. On desktop, tools usually treat the cursor moving toward the browser tabs or the address bar as a sign of leaving. Phones have no cursor, so the intention to leave is inferred from other signals. Find out which behavior your tool looks at on mobile and test the survey on your own phone; if the popup opens while the visitor is still reading the page, change the setting, because an option selected at that moment may simply be a way to close the popup.
Do not show the survey to someone who has just landed on the page. A visitor who leaves within a few seconds has not had a chance to consider the product, and their answer will not tell you much. Once a visitor has answered the survey or closed the popup, we recommend not showing them the same survey again for a few weeks. If a spin-to-win wheel, an invitation to sign up and an exit survey open one after another during the same visit, choose one of them; do not expect a survey answer from a visitor who has just closed three popups in a row. Each week, also note how many times the popup opened and how many answers and dismissals it received; if the response rate drops suddenly, first check when the popup opens.
How much of the screen the popup covers matters as well. According to Google Search Central documentation, interstitials that obscure the content can frustrate users and erode their trust in the site. The document advises against covering the entire page and recommends banners that take up only a small fraction of the screen; that advice is not limited to phones. On a small screen, however, even a medium-sized popup hides most of the content. On phones, design the survey as a short box at the bottom of the screen with a clearly visible close button.
Ask one question and write the options in the visitor’s words
Start with a single question. The visitor is already on the way out; do not expect them to fill in a second screen as well. Ask the question with ready-made options, and end the list with an “Other” option that opens a text field. Show a second question only to visitors who select a particular option, to find out more about that answer; when the popup first opens, there should still be only one question on screen. For example, if a visitor on the search page says “I couldn’t find what I was looking for,” following up with “What were you looking for?” gives you information you can use directly to adjust your search and filters.
To start with, we recommend no more than six options on desktop and no more than four on phones. On a phone, all the options should be visible without scrolling; a visitor may close a list that requires scrolling without reading it. Check the options against these criteria as you write them:
- Each option should cover a single topic. A combined option such as “Price and shipping are too expensive” will not tell you which one to fix when answers come in.
- Write the options in the words visitors use. “I got stuck at the payment step” is easier to pick than “The checkout flow is complicated.”
- Always include an “I was just browsing” option. If you do not separate out the visitors who were not ready to buy, you will read the other reasons’ shares as higher than they really are.
- Do not ask a leading question. “Did you find our prices too high?” suggests an answer; “What stopped you?” does not.
Make the question single-choice, so that visitors can select only one option. With multiple choice, a visitor can select several reasons at once, which makes it harder to tell which reason comes first. Read the answers you already have before adding more questions.
Do not reward answers or ask for contact details
Offering a discount code to visitors who answer the survey may increase the number of answers, but it can also change who answers. A visitor who clicks on the popup for the code may select one of the options without reading them, and the share of coupon hunters among respondents may grow. If you want to make an offer to visitors on their way out, use a separate popup for it and do not show both popups during the same visit.
Adding an email or phone field to the survey form also muddles its purpose. Someone ready to answer a short question may not be ready to share personal information. Moreover, an email address or phone number is personal data, because it relates to an identified or identifiable person. Article 10 of Türkiye’s Law No. 6698 on the Protection of Personal Data (KVKK) requires the data controller, or a person it authorizes, to inform the data subject when collecting personal data. The notice covers five points: who the data controller is, along with its representative if it has one; why the data is processed; to whom the data may be transferred and why; how the data is collected and on what legal basis; and the data subject’s rights set out in Article 11 of the Law. If you want to stay in touch with visitors, ask for contact details with a clear invitation and a privacy notice, in a separate popup such as a lead capture popup; keep only the options and a short text field inside the survey.
A name, phone number or order number typed into the “Other” box can also be personal data. Do not add such information to your marketing list; when you read the answer, note only the reason for leaving. Nor should you publish negative comments left in the survey as customer reviews; the visitor did not write their answer expecting it to be published.
Split the answers by page type and device
Do not make a decision based on the first week’s answers. When there are only a few dozen answers, one option standing out may be a coincidence. We recommend waiting until about a hundred answers have built up for the same page type; if you have few visitors, that can take several weeks. Decide from the start who will read the answers each week and whom they will pass the findings on to; then read the answers in four steps.
- Put the “I was just browsing” answers to one side and work out the percentages from the remaining answers. The reasons you need to fix are the ones that stop visitors who intend to buy.
- Split the answers into desktop and mobile; if your tool shows traffic sources, split them by source as well. If the delivery date sits next to the price on desktop but at the bottom of the page on phones, expect most of the complaints to come from phones; if you combine the answers, you will not see that difference. When a visitor who arrived from a price comparison site or a discount ad answers “The shipping fee is higher than I expected,” it may also mean something different from the same answer given by a customer who came by searching for your brand.
- Read the entries in the “Other” box every week and give each one a short tag. If the same tag keeps coming up, give it an option of its own. If answers in the “Other” box approach a quarter of the total, review your options; visitors’ real reasons may not be on the list.
- Tag answers that arrive during campaign and sale weeks separately; the profile of the visitors coming to your site may change during those periods.
Keep one limitation in mind as you read the answers: only visitors who choose to respond take part in the survey. The results do not give you the exact breakdown of all the visitors who leave; some groups of visitors may be more willing to answer than others. Even so, you can use the results to set priorities among the reasons.
Problems that customers run into with the product or the shipping after completing their order are beyond the reach of an exit survey. You can measure them with a WhatsApp feedback survey sent to customers after their order is delivered: the customer first selects a score from a preset list and then, if they wish, describes their experience in a few sentences. When you put the two sources side by side, you can tell the reasons that stop visitors at the door apart from those that leave customers dissatisfied after delivery.
Example: four weeks of cart page answers
The figures below belong to a hypothetical home textiles store and are only meant to show how to read the answers. Over four weeks, the store collects 240 answers in its desktop cart survey: 72 people say “I was just browsing,” 66 “The shipping fee is higher than I expected,” 38 “The delivery date is too late” and 24 “The payment option I want isn’t available”; another 40 write in the “Other” box.
Setting aside the 72 “I was just browsing” answers leaves 168 answers in the calculation. The shipping fee complaint makes up about 39% of these, the delivery date 23% and the payment option 14%. At 24%, “Other” is close to the review threshold. When the store reads those 40 entries, it sees that 25 of them say “the discount code wasn’t accepted.” It first investigates and fixes the bug in the code field; it adds a “My code didn’t work” option only after the period in which it measures the effect of the shipping change is over. For the shipping complaint, it looks at the share of carts abandoned just short of the free shipping threshold, and decides to display the threshold on the cart page, along with how much is left to reach it. If the store counts the answers to its phone cart survey separately and the delivery complaint’s share comes out clearly higher than on desktop, the problem may be that the delivery date is not visible in the mobile cart.
Tie each reason for leaving to a single site change
For each main reason that emerges from the survey, assign one owner and one change. If you change five things at once, you cannot tell which change worked. For common reasons, these are the first steps to try:
- If the shipping fee comes as a surprise, show the free shipping threshold and the shipping fee on the product and cart pages, before checkout. Before you decide whether to send a discount code to abandoned carts, calculate when a discount code remains profitable.
- If the delivery date is seen as too late, put the estimated dispatch date close to the price. If your delivery time really is long, do not hide it; give the reason. For example, if the product is made to order, say so on the product page.
- If the problem is a missing payment option or a lack of trust, make the payment methods you accept and your return policy visible in the cart. To find out which method visitors are looking for, you can ask those who say “The payment option I want isn’t available” a follow-up question: “Which payment method were you looking for?” Installments, cash on delivery and bank transfer each call for a different arrangement. If mandatory account creation draws complaints, consider letting customers order without an account.
- If product information is found insufficient, complete the size charts, measurements and images, starting with the products that receive the most answers. Add the same information to your help page as well if it is missing there.
- There is nothing to fix on the site for visitors who say “I was just browsing.” Among them, there may be customers who added a product to their cart and gave you permission to contact them; you reach them independently of the survey, with a reminder sent to customers who abandon their cart. The WhatsApp cart recovery module sends that reminder only to shoppers who have opted in. For the timing of the message, see our article on when to send an abandoned cart message.
Track the effect of the change with the same question
After making a change, leave the survey question and options as they are. On the same page type, compare two periods of equal length, one before and one after the change, and look at the share of answers that give the reason you addressed. Alongside the share, also divide the number of visitors who chose that reason by the number of visitors who saw that page in the same period, since the share can also fall because another reason has grown. Exclude from the comparison any weeks in which you added an option or fixed another bug.
Suppose the example store starts showing the shipping threshold in the cart. If the shipping complaint then drops markedly, both as a share of the answers and as a ratio to the number of visitors who saw the cart page, the change has most likely hit the right spot. The week in which the store fixed the code bug should also stay outside both periods. If both the share and the ratio stay put, the problem may lie less in the visibility of the information than in the fee itself; the next step is to review the threshold or the shipping fee. Do not decide based on a difference of a few points; check whether a difference in the same direction continues into the next period.
Read the survey figures together with your analytics data. If the rate of visitors moving from the cart to checkout also rises as the shipping complaint falls, the two sources are telling you the same thing. If one of the two periods you are comparing coincides with a campaign or sale week, do not attribute the difference to the change; the campaign weeks you tagged separately when reading the answers come in handy here. If other reasons gradually take over the top of the list, you may have dealt with the problems that used to rank first; if the same reason stays first for months, the change you made may not have fixed it. Pushouse’s exit-intent popup page also includes sample questions and options for desktop and mobile visitors.
