Almost never on the first visit, and never in the first few seconds. A box that covers the page before the visitor has read a line is asking for something before you have given anything, and most people close it without reading what it said.
There is a narrow set of cases where a pop-up earns its place. The rest are there because a plugin made them easy to add.
What is the pop-up actually asking for?
An email address, usually. Sometimes a discount code in exchange for one. Sometimes a survey, an app download, a chat greeting, or a notice that you are closed on Monday.
Look at it from the visitor's side. They typed a question into Google, tapped your page, and before it finished loading a box slid over it asking for their email. They do not know yet whether you sell what they need, whether you serve their town, or whether your prices are in their range. You have asked for the thing customers give last, first.
The close button is what they tap. On a phone it is a small cross in a corner, sometimes not where the thumb expects it, and a mis-tap lands on the form instead. That is the whole experience for most people. Arrive, dismiss, start reading with a little less patience than they had a moment ago.
What does Google say about it?
Google's Search Central documentation has a page on intrusive interstitials and dialogs, and it is direct. It tells site owners not to obscure the entire page with interstitials, and says that dialogs which get in the way of the content "may lead to poor search performance". What it recommends instead is a banner that takes up only a small fraction of the screen, so that visitors and search engines can reach the content as soon as they land.
Two details on that page matter. Mandatory interstitials, such as an age gate or a notice you are legally required to show, are exempt. And the concern is what a visitor arriving from search sees first. A box that appears after someone has been reading for a while is a different case from one that meets them at the door.
So the plugin that fires a full-screen "Get 10% off" on every page, on every phone, on arrival, is the exact pattern Google's own guidance describes as the thing to avoid.
Does the discount pop-up on a store work?
It collects email addresses. That is not the same as making sales.
Think about who submits it. Some are people who were going to buy anyway, and now they buy for ten percent less. Some are people who were never going to buy, and now you have their address and they have a code they will not use. The group you wanted, the undecided visitor who needed one more reason, is in there too, but the plugin's dashboard counts all three as the same win.
There is a slower cost. A store that hands out a code on arrival teaches repeat customers to wait for it, open the site in a private window, or search for "your brand discount code" before every order. You have turned a first-purchase nudge into a permanent price cut.
The offer works better where the decision happens. On the product page, next to the price. At the cart, for a first order, when someone has already shown they want the thing. A box on the homepage is the wrong room for it.
When does a pop-up earn its place?
When it appears after the visitor has done something, says one thing, and gets out of the way.
After a real signal. Someone has read most of a long article, spent a few minutes on a service page, or added an item to the cart. A small box at that point, offering something related to what they were just doing, is a fair exchange. The same box on arrival is not.
On the way out, on desktop. Exit-intent boxes trigger when the cursor heads for the back button or the address bar. That only exists on a computer. On a phone there is no cursor, so "exit intent" quietly becomes a timer or a scroll trigger, which is the arrival pop-up again with a delay.
When it changes what the visitor would do. Closed for the long weekend. Last day to order for delivery before a holiday. A price change that takes effect next month. One line, one dismiss, and the site remembers that they closed it.
When you are required to show it. A legal notice or an age check is not a marketing decision, and Google's guidance treats it separately.
Whatever the trigger, the rules are the same. Once per visit, not once per page. Small, not full screen. A close button a thumb can hit. Remembered once dismissed. And never on the page someone lands on from search.
What should you use instead?
The same message, placed where it belongs.
A slim bar across the top or bottom of the page for the announcement, the delivery cutoff, or the offer. This is the "small fraction of the screen" banner in Google's own recommendation, and it stays readable without covering anything.
An inline block inside the page, where the reader is already looking. The newsletter box at the end of the article, for the person who just finished it. The first-order offer beside the price. The "book a call" prompt after the section that explains what a call involves.
A footer signup for the visitor who scrolled to the bottom, which is a visitor who already decided they liked you enough to keep going.
A sticky contact bar on mobile if what you actually want is the enquiry, not the email. A call button and a message button that stay within reach do more than a box asking for a newsletter subscription. What that bar should link to is covered in traffic but no leads.
If the pop-up is a chat greeting, that is a separate question with its own trap, and it is answered in should you add live chat to your website.
What does the pop-up cost beyond the subscription?
Another third-party script on every page, and another thing that breaks the mobile layout.
Pop-up tools load from their own servers, in addition to the analytics, the pixel and whatever else is already on the page. Each one is a small delay on a phone, and together they are often why a site that should feel quick does not. That trade is explained in why your website is slow.
The layout problem is more visible. A box designed on a wide monitor lands on a narrow phone with the close button off the edge, or behind the browser's own bar, or under a cookie notice. The visitor cannot dismiss it, cannot scroll behind it, and leaves. If you have never opened your own site on a phone from a Google search result, that is the test, and the rest of it is in is your website mobile-friendly.
There is a craft point too. A pop-up is a dialog, and a dialog that cannot be closed from the keyboard, or that a screen reader does not announce, locks some visitors out of the page entirely. Done properly, it takes focus when it opens, closes on Escape, and returns focus to where the reader was. Most plugins do not bother, and the site owner never finds out.
How do you tell whether yours is helping?
Measure sales and enquiries, not signups.
The plugin will show you the number of emails it collected, and that number always looks like progress. It will not show you how many people closed the box, or how many left the site without closing it. Those two are the numbers that matter, and they are missing from the dashboard.
The honest test is to turn it off for two weeks and compare orders or enquiries against the two weeks before. If the difference is nothing, the pop-up was collecting addresses, not customers. If enquiries go up with it off, you have your answer. What to look at, and where, is in how to tell if your website is working.
What does this cost to fix?
Very little. Removing a pop-up, moving the offer into the page, and adding a slim bar for announcements is a short job. On a Shopify store the same move, plus a first-order discount at the cart instead of the door, is standard work in any build we do.
If the site around it is the problem, our marketing sites and landing pages run from $1,200 to $8,000, and store and B2B builds start at $2,900. The ranges are on pricing, how a marketing site gets scoped is on web design, and store builds are on e-commerce.
If you want to know what your visitors see in the first five seconds on a phone, and whether a box is getting in their way, send us the URL for a free teardown. We will tell you what is covering the page, what it is asking for, and where that ask should sit instead, whether or not you work with us.