How to get more people to say yes to push notifications

The question I get most about opt-in is what a good opt-in rate is. I understand why. It is the first number on the dashboard and it is easy to compare.

It is also the wrong place to start. A prompt that pops up on every page load can produce a big list quickly. Then half of it never clicks anything, the browser starts revoking permissions, and you are paying to keep a list that mostly exists on paper. The better question is how many of the people who said yes are still reachable, and still interested, a month later.

This guide is about getting more of those people.

Why the page-load prompt fails

The browser's permission dialog is a blunt instrument. It has no context, it can't explain anything, and the visitor has usually been on your site for about a second when it appears.

The 2025 Web Almanac PWA chapter, from an HTTP Archive crawl in July 2025, reports that desktop users ignore notification prompts 78% of the time, and that acceptance on mobile is 16%. That is prompt behavior across a crawl, not your site, but it tells you what you are up against when you ask strangers before they have read anything.

There is a second cost that is easy to miss. When Chrome began rate limiting the Push API in January 2026, one of the daily factors it named was permission prompts per time on site. So the prompt that greets every visitor on arrival is being counted, mostly by people who were never going to say yes.

And a visitor who clicks block on the native dialog is very hard to ask again. The browser remembers. You have spent your one real chance on someone who didn't know you yet.

Ask in two steps

The pattern that works is a soft prompt first, then the browser dialog.

Step one is yours. A small box or banner in your own design and your own words: what you will send, roughly how often, and two buttons. If the visitor says not now, nothing happens and you can ask again another day.

Step two is the browser's. Only after someone clicks yes on your prompt do you trigger the native permission dialog. Now the person expects it, knows why it is there, and has already made the decision once.

The soft prompt costs you nothing when it gets a no. The native dialog does.

When to ask

Timing matters more than wording. The visitors worth asking are the ones who have already shown some interest.

Reasonable triggers to test:

• After a second pageview in the same session.
• After the visitor has scrolled most of the way through an article.
• After a specific action, like finishing a recipe, reading a match report or checking a price.
• On a returning visit rather than the first one.

Each of those filters out people who were never going to say yes and would have cost you a prompt. Your list grows slower and is worth much more.

One thing to avoid: a prompt that triggers on a timer alone. Ten seconds on a page says very little about whether someone is reading.

What to write

Keep the soft prompt short. One sentence about what they get, one line about frequency if you can say it honestly, and two clear buttons.

Some patterns that tend to work better than "Allow notifications?":

• Say the content, not the channel. "Get the match result the moment it's final" beats "Subscribe to notifications".
• Say how often. "A few times a week, never at night" removes the biggest fear people have.
• Match the page. A prompt on a recipe page can mention recipes. The same generic prompt everywhere wastes context you already have.
• Make no a real option. "Not now" works better than a guilt line. People remember being pressured.

Avoid fake urgency, fake system dialogs and anything designed to look like the browser's own prompt. Apart from being dishonest, it teaches people that your site tricks them, which is the opposite of what you need from a permission.

What to measure instead of the opt-in rate

The raw opt-in rate rewards asking more people more aggressively. These numbers reward asking the right people:

If the soft prompt yes rate is low, change the offer or the timing. If people say yes and then unsubscribe within days, the prompt promised something the sends don't deliver.

The welcome message is part of the opt-in

The first notification someone gets should arrive right after they subscribe. It confirms that it worked, repeats what they will get, and gives them something useful to tap. It is also the moment people decide whether this was a mistake.

A welcome that says "Thanks for subscribing!" and nothing else is a wasted send. A welcome that links to something they would actually want, and says how often you will be in touch, sets up everything after it.

Limitations

• Prompt behavior differs by browser and device. On iPhone and iPad, web push works for sites added to the Home Screen, so a prompt on a regular Safari visit on iOS has nothing to ask for.
• There is no reliable public benchmark for opt-in rates by industry. The vendor posts that publish them rarely describe their sample. Compare your site against itself over time.
• A higher opt-in rate is not automatically better. If it rises and the 30-day reachable share falls, you are collecting people who will leave.

How EngageNudge handles opt-in

EngageNudge shows a soft prompt in your words before the browser dialog, never the other way around. You choose when it appears: after a number of pageviews, after a scroll depth, or after a delay. A welcome Nudge goes out right after subscription, with an optional short follow-up.

After that, every send is checked per person against daily and weekly caps, quiet hours and recent ignores, so the people who said yes don't get worn out by the sends that follow.