Exit Intent Popup
An exit intent popup is an overlay triggered the moment a visitor's behavior signals they are about to leave a page, presenting a final offer to keep them engaged.
Key takeaways
- Detection depends on pointer velocity and distance from the viewport's top edge.
- Loose thresholds interrupt engaged visitors reaching for bookmarks or other tabs.
- The exit offer must be lighter than the request the visitor just declined.
- Touch devices provide no reliable cursor signal, so mobile proxies misfire or break navigation.
- An overlay arrives after the decision to leave, so it cannot fix price or fit objections.
In depth
Exit intent is a detection rule rather than a design. A listener watches the pointer's position and upward velocity, and when the cursor crosses the top edge of the document toward the address bar or tab strip, the overlay renders. Two thresholds decide how it behaves: how close to the edge the pointer must be and how fast it has to be travelling. A session flag then stops it firing again, so each visitor gets at most one attempt as they leave.
Sensitivity is the whole trade-off. Loosen the thresholds and the popup fires more often, including on people reaching for a bookmark, switching to another tab or moving toward a second monitor, and those interruptions land on visitors who were still engaged. Tighten them and genuine leavers escape before the overlay appears. What the popup offers matters just as much: the visitor already declined the page's main request, so repeating it converts almost nobody. The page it fires on shapes the result too, since a pricing page leaves a specific unanswered question.
In practice the exit offer is lighter than the one just refused. A pricing page that asked for a demo can offer a comparison sheet or a short fit assessment; a scorecard suits this moment because it promises an answer rather than asking for an address, and the contact fields arrive after the score exists. The copy should acknowledge the departure in one line, keep the form to a single field, and let the visitor close it without hunting for the control.
The technique has no dependable equivalent on touch screens, where there is no cursor to track. Proxies such as rapid upward scrolling, an idle period or a manipulated history entry either misfire constantly or interfere with the back button, which is worse than showing nothing. Exit intent also cannot answer the reason someone is leaving: if the price, the fit or the trust was wrong, an overlay arrives long after that judgement was made. Anyone who dismissed it once should be excluded permanently, because a second attempt only repeats a question already answered.
Example in practice
How to measure it
Start with trigger rate, the share of sessions where the overlay fires, and conversion per trigger. A trigger rate that looks high for the page's exit rate means the thresholds are catching people who were not leaving. As a rough false-positive check, count the sessions where the popup fired and the visitor then kept browsing for a while; those interruptions cost engagement without recovering anyone.
The important number is incremental. Hold back a share of traffic that never sees the exit popup, then compare total conversions per session between the two groups rather than counting popup submissions as recovered visitors. Some of those people would have returned or converted later anyway. Split results by device and page type, since pricing pages and blog articles usually behave nothing alike.
Common mistakes
The first mistake is repeating the ask. A visitor who did not book a demo is shown a modal asking them to book a demo, or a discount that means nothing to a buyer whose objection was fit rather than price. Change the shape of the request instead: offer something that answers the question they were left with, keep it to one step, and make it useful even to someone who never becomes a customer.
The second is careless detection. Firing on any mouseleave event catches every visitor who glances at another window, and pushing a fake history entry so the back button opens a popup breaks the one control people trust. Require upward movement toward the top edge, ignore exits at the left and right, set a session flag after the first display, and never intercept a deliberate navigation away.
Frequently asked questions
Does an exit intent popup hurt user experience?
Not when configured well, because it only appears for visitors who are already leaving rather than interrupting active reading. Keep it single-purpose, easy to dismiss, and capped to one impression per session to avoid annoyance.
How does exit intent detection actually work?
A script tracks the pointer and fires when it moves upward past a threshold near the top of the viewport, or when the document receives a mouseleave event in that direction. Speed matters as much as position, since a slow drift is not a departure. Downward or sideways exits are ignored because they usually mean the visitor is reaching for something else.
Does exit intent work on mobile?
Not reliably, because there is no cursor and no equivalent gesture that clearly signals leaving. Available proxies are weak: fast upward scrolling also happens while skimming, idle time also means reading, and history manipulation to catch the back button breaks the control people rely on. Most teams disable exit intent on touch devices and use a scroll-triggered offer instead.
What offer converts best in an exit intent popup?
Something lighter and more immediate than the request the visitor just declined. A comparison sheet, a short assessment or an answer to the question the page raised all work, because they cost the visitor one step. Discounts convert in retail contexts but rarely move a business buyer whose objection was fit. Whatever you choose, it must be deliverable in a single interaction.
How do I know exit popup conversions are incremental?
Run a holdout. Keep a share of traffic that never sees the overlay, then compare conversions per session across the whole page for both groups. Counting popup submissions as recovered visitors assumes every one of them was lost, which is not true: some would have returned directly, and others were interrupted while still engaged.
Should an exit intent popup fire during checkout?
Only with a very specific reason, such as offering help or saving the cart, and never as a generic newsletter prompt. A checkout is a task, and interrupting it can lose an order that was about to complete. If you do use one, target the step where abandonment actually concentrates and measure completed orders rather than popup submissions.
Why does my exit popup fire when visitors are not leaving?
The thresholds are too loose, so the script counts any pointer movement toward the top of the window as a departure. Tighten the required speed and the distance from the edge, ignore exits through the sides, and add a minimum time on page before it can arm. Check the share of triggered sessions where the visitor continued browsing afterwards.