GuideEight rules, and three times the answer is no popup at all.
The usability literature is critical of popups and it has a point. This is written by a company that sells them, so it starts by taking the criticism seriously rather than waving it away.
The eight.
Ordered by how much damage getting them wrong does.
Never fire on arrival
A visitor who has seen nothing has no reason to want anything. Waiting for a scroll depth, a delay or an exit signal costs almost no reach and changes who you are interrupting from everyone to people who stayed.
Cap how often one person sees it
The second popup in a session is noise; the third is a reason to leave. This single setting does more for perceived quality than any amount of design work.
Never show it to someone who already signed up
Asking a subscriber for an address they gave you last week is the most common self-inflicted wound in this category, and it corrupts your conversion numbers as well as their experience.
Make the close control obvious and large
A small ✕, a low-contrast ✕, or a dismissal that requires clicking a self-deprecating “No thanks, I hate savings” link are all dark patterns. They win a few signups and lose trust that was worth more.
Say something specific
“Join our newsletter” asks for an address in exchange for advertising. A named offer with a number converts several times better on identical traffic, which makes vagueness the most expensive thing on the page.
Ask for one field
Every additional field costs completions. Name, birthday and preferences can all be collected later from someone who already said yes — and who is far likelier to answer.
Treat mobile as a different problem
No cursor means no true exit intent, the keyboard can take half the viewport, and Google's intrusive interstitial guidance applies to mobile arrivals from search but not to desktop.
Change one thing at a time
Redesigning the offer, the timing and the copy together tells you the result changed but not why. Most popup programs plateau because nobody can attribute the last four wins.
Three times the right number is zero.
A popup vendor is supposed to leave this section out.
In checkout
A visitor with their card out is not a lead to capture. Interrupting payment to ask for an email trades a confirmed sale for an address. Fire on the cart, never on the checkout.
When the page has nothing to offer
If the honest content of the popup is “please stay”, the popup is not the problem being solved. It will not fix a page nobody wants to read, and it will make the bounce feel deserved.
On a site with barely any traffic
At a few hundred sessions a month, a popup produces numbers too small to learn anything from, and you will tune it on noise. Get traffic first; the capture layer is worth almost nothing without it.
Rules one to three are settings, not discipline.
The three rules that matter most are the ones a tool should make hard to get wrong. In BetterPopup that means 7 trigger conditions so nothing has to fire on arrival, a frequency cap per campaign, and 6 audience conditions — the first of which is whether this person has already signed up.
Popup practice, answered plainly.
Starting with whether they are bad for users.
Usability research has been critical of them for a long time, and the criticism is largely fair — they interrupt, they obscure content, and they are frequently implemented with dark patterns around the close control. That is a real cost. The reason they persist is that a well-timed one, capped properly and not shown to existing subscribers, converts far better than any static form on the page. Both things are true, and a vendor telling you only the second half is not being straight with you.
After the visitor has done something suggesting interest — reached a scroll depth, spent time on the page, added something to the cart, or shown a signal that they are leaving. Firing on arrival is the single most common mistake and the one that generates most of the hostility toward popups as a format.
Rarely enough that they do not notice a pattern. Once per session is a reasonable default, and once per visitor per campaign is safer still. There is no universal number — it depends on how often people return — but the failure direction is almost always too often, not too rarely.
In one configuration: content obscured on mobile immediately after a click from search results, which is what Google's intrusive interstitial guidance addresses. A popup triggered by a scroll, a delay or an exit signal does not match that pattern, and desktop is not covered by it at all. Cookie and legally required notices are treated separately.
We will not publish a number, and you should be sceptical of anyone who does. Every benchmark in circulation is a vendor summarising its own customers, often its best campaigns, on traffic that looks nothing like yours. Measure your own rate as a percentage of sessions, split mobile from desktop, change one variable, and compare against yourself.
Start with one and make it work. The most valuable second campaign is usually not another offer on the same visitor but a different offer at a different moment — typically a cart-stage popup alongside a first-visit one. Beyond two or three concurrent campaigns, most sites are competing with themselves and cannot tell which one produced a result.