It only fires on desktop, catches people already leaving, and is judged against a low bar. Measure what it actually produces before assuming it is free.

What it does and when

Exit intent watches the cursor and triggers when it moves rapidly toward the top of the window, which is where the close button and the address bar are.

That is a guess about intention, and it is a reasonable one for a desktop visitor.

On a phone there is no cursor, so it cannot work the same way, and implementations that claim to are guessing from scroll direction or a timer instead.

Which means a small business site, where most traffic is mobile, is running the feature for a minority of visitors.

The argument for it

Those are genuine, and the first is the strongest: an overlay that appears mid-read is a cost, and one that appears as somebody leaves has almost nothing to lose.

The argument against

Which is mostly about what it actually achieves rather than what it costs.

Somebody heading for the close button has already decided, and a panel is unlikely to reverse that.

The people it does capture are frequently the least committed, which shows up later as a subscriber who never opens anything.

It fires wrongly a good deal: somebody reaching for a bookmark, another tab, or the back button gets it, and some of those were not leaving.

And it adds code to every page for a feature that fires on a fraction of desktop sessions.

None of that makes it wrong, and it does mean the benefit needs measuring rather than assuming.

Measure what it produces

The part that settles it, and it is rarely done.

Record three numbers over a month: how many times it displayed, how many people acted on it, and how many of those became anything.

The second divided by the first is usually a small percentage, which is fine and expected.

The third is the one that decides. A hundred captures producing no enquiries is a hundred addresses you will not use.

Compare it against the same period without it running, since the honest question is whether total enquiries changed rather than whether the panel collected anything.

Most tools report the first two and none report the third, which is why it has to be traced deliberately.

A worked example

A business added an exit panel offering a guide in exchange for an email address.

Over three months it displayed about two thousand times, collected sixty-one addresses, and produced two enquiries.

The sixty-one were added to a mailing list, of which most never opened anything, which lowered the open rate for the list as a whole.

They replaced it with a different panel offering a callback, with a phone field and nothing else.

That collected far fewer, around a dozen, and produced five enquiries, because it asked for the thing somebody leaving might still want.

The lesson was about what was offered rather than about the mechanism.

Offer the right thing

Since most exit panels offer whatever the tool suggested.

A newsletter signup is the default and the weakest, because somebody leaving your site has no interest in hearing from you monthly.

A discount works where you sell something directly and the person was comparing on price.

A callback or a question works for a service business, because it is a smaller step than the enquiry they just declined.

And for a page where somebody was clearly looking for something specific, an offer of that specific thing beats anything generic.

Ask what the person on this page was trying to do and offer a cheaper version of it.

The rules if you use one

Practical constraints that keep it from being a nuisance.

Once per visitor, remembered, rather than on every page they nearly leave.

A large obvious close button, since a panel that is hard to dismiss converts a departure into an annoyance.

One field, since somebody leaving will not complete four.

Never on mobile, where the trigger is a guess and the interruption is worse.

And not on the contact page, where the person may be reaching for something legitimate and does not need an offer.

The alternative that always works

Worth considering before installing anything.

A page ending with a clear prompt, and a visible phone number, catches the same people earlier and without a panel.

Somebody who reached the bottom and found nothing to do is a more addressable problem than somebody heading for the close button.

Fixing the page is permanent, applies on every device, and costs no code.

Where a business has both a weak page ending and an exit panel, the panel is compensating for the page and the page is the thing to fix.

Watch what it does to the rest

A cost that is easy to miss because it appears somewhere else.

Addresses collected by an exit panel are frequently the least engaged on a list, so adding them lowers the open rate and raises the unsubscribe rate for everybody.

That affects how mail providers treat your domain, which affects the messages you send to people who actually wanted them.

Keep them as a separate group for the first few sends and watch how they behave before merging them into the main list.

Where they behave badly, that is the answer about the panel, arriving from a direction nobody was looking.

The counter-case

It genuinely works for some.

Sites with a real offer, high desktop traffic, and something worth exchanging an address for do collect useful volume this way.

It is also among the least intrusive overlay patterns, and objecting to it as strongly as to one that interrupts a reader is disproportionate.

And it is easy to test and easy to remove, which is more than can be said for most changes.

Measure it against enquiries rather than captures, offer something a leaving visitor would want, and fix the page ending first.

What to do

  1. Fix the page ending before adding a panel.
  2. Record displays, captures, and enquiries.
  3. Judge it on the third number.
  4. Offer something a leaving visitor wants.
  5. One field, once per visitor.
  6. Never on mobile or the contact page.
  7. Remove it if enquiries did not move.

Step three is where these are usually misjudged, since capture counts look encouraging and say nothing about whether anything came of them.

The mobile version of this problem is covered in pop-ups on a phone.

Frequently asked questions

How does exit intent work?

It watches the cursor and fires when it moves toward the top of the window. There is no cursor on a phone, so mobile implementations guess from scroll or a timer instead.

What is the argument for it?

The visitor was leaving anyway, so it does not interrupt somebody reading, and any capture is one that was otherwise lost. That is genuine as far as it goes.

What is the argument against?

Somebody heading for the close button has decided, the people captured are the least committed, and it fires wrongly on bookmarks, tabs, and the back button.

How should I measure it?

Three numbers: displays, captures, and enquiries resulting. Most tools report the first two and none report the third, which is the one that decides.

What should the panel offer?

Something a leaving visitor would still want. A newsletter is the default and the weakest; a callback or an answer to a specific question works better for a service business.

What is the alternative?

Fixing the page ending. Somebody who reached the bottom and found nothing to do is a more addressable problem, and the fix applies on every device.

West Coast Media Solutions Inc. provides web design, web development, hosting, digital marketing, and business consulting to organisations across Canada, drawing on more than twenty-five years in the field.

Exit panel collecting addresses?

Trace how many became enquiries, not how many were captured. Those are different numbers and only one matters.

Start a Conversation