Most blog traffic is one-and-done. Someone searches, lands, reads, leaves, and never returns. You can have a hundred thousand monthly visitors and almost no audience.
Push notifications are one of the few ways to convert a one-time reader into a returning one without asking for an email address. No form to fill in, no inbox to compete with, no double opt-in. One click and you can reach them again.
Done badly, though, push is worse than useless — it teaches people to block you, permanently, with no way back. Here is what separates the two.
The numbers to expect
Before anything else, calibrate. Vague promises of “10x traffic” are how this channel gets a bad name.
Rough figures for a content site with a well-timed prompt:
| Metric | Realistic range |
|---|---|
| Opt-in rate (prompt shown → subscribed) | 5–12% |
| Click rate per notification | 3–10% |
| List decay per year, no unsubscribes | Substantial — plan to keep growing it |
Run the arithmetic on your own numbers before you invest much effort. A site with 30,000 monthly visitors, a 7% opt-in rate and two notifications a week at a 6% click rate is adding a modest but real slice of repeat traffic — and it compounds, because the list grows every month while the sending cost stays zero.
That is the honest shape of it: not transformative on its own, genuinely useful stacked with everything else you do.
Getting the prompt right
This is where most of the outcome is decided.
Never fire the native browser prompt on page load. A visitor who has read nothing has no reason to want notifications from you. They click Block, and that Block is permanent — you cannot ask again, ever, unless they go into browser settings and reverse it. You do not get a second chance with that person.
Use a soft ask first. Show your own box — your wording, your branding — before triggering the browser’s prompt. If they say no to yours, nothing is lost; you can ask again in a week. Only the people who say yes to your box ever see the browser dialog, so your Block rate collapses.
Ask after they have got value. Good triggers, roughly in order of how well they tend to work:
- After 50–60% scroll depth on an article. They have read most of it; they like your writing.
- After 30–45 seconds on the page.
- On exit intent, desktop only.
- On click of a “get updates” button you place yourself.
Say what they are signing up for. “Get updates from Example” is weak. “New tutorials on WordPress and PHP — about two a week” tells them the topic and the frequency, which is what they actually want to know. Frequency in the prompt reduces unsubscribes later.
What to send
The rule: a notification should be worth the interruption to the person receiving it, not to you.
Works well:
- New post announcements, when your posts are genuinely worth reading
- Time-sensitive things — a deadline, a result, a live event
- A follow-up to something they already engaged with
- Genuinely good offers, used sparingly
Works badly:
- Every post on a high-volume site. If you publish five times a day, do not push five times a day. Pick the best one.
- Re-pushing old content with a new headline
- Anything that reads like clickbait. It gets a good click rate once and a terrible one forever after.
- Notifications sent at 2am local time
Frequency
This is the part people get wrong most consistently.
Start at two to three a week. Watch what happens to your click rate. If it holds steady, you have room. If it drops, you are sending too often.
The failure mode is slow and easy to miss: your click rate falls a little each week, so each notification earns less, so you send more to compensate, so it falls further. By the time you notice, you have trained your entire list to ignore you — and unlike email, where people just stop opening, push users actively block, which removes them permanently.
If you have a large list and something genuinely valuable to send more often, use frequency capping (EasyPush Pro has this): a hard limit of one notification per subscriber per day regardless of how many campaigns fire. Transactional messages — an order update, a cart reminder — should bypass the cap, because those are things the person actually asked for.
Writing the notification
You have about 60 characters of title and 120 of body before things get truncated, and that varies by browser and OS. Write to the shorter end.
Title: the hook. Not the article’s headline verbatim — the headline was written for search, the notification is written for someone glancing at a corner of their screen.
Body: the reason to tap. Specific beats vague. “New guide” is nothing. “Why your WordPress site is slow even on good hosting” is something.
Image: a wide hero image measurably lifts click rate on desktop Chrome and Android. It is ignored elsewhere, so it costs nothing to include.
Buttons: two action buttons, each with its own link. Useful for “Read now / Save for later” patterns. Not supported everywhere, so never put anything essential in them.
Measuring it honestly
One thing to be careful about in whatever tool you use: check what the click rate is measured against.
If it is clicks divided by sent, the number is inflated. “Sent” means the push service accepted the message — not that it reached a device, and not that a human saw it. Devices that are off, browsers that never opened, subscriptions that have quietly died all count as “sent”.
The honest denominator is displayed — notifications that actually rendered on a screen. That is the number of real opportunities to click. It is always smaller, so the resulting CTR is always lower, and it is the only one that tells you whether your copy is working.
If your push tool only reports against sent, mentally discount the figure.
A practical starting plan
Week 1. Install, set the prompt to fire at 50% scroll, write a clear opt-in message that names your topic and frequency. Send nothing.
Week 2. You should have a few hundred subscribers. Send one notification for your best post of the week. Note the click rate.
Weeks 3–6. Two a week. Vary the style — question, statement, number. Watch which shape performs.
Week 7 onward. You now have your own baseline, which is worth more than any benchmark in this post. Adjust frequency against your own click rate, not against what someone on the internet said.
Add a welcome notification once you have the rhythm: a single message a minute or two after someone subscribes, confirming what they signed up for. It costs you nothing and it measurably improves how engaged that subscriber is a month later.
What to avoid
I have written the failure modes up separately in the mistakes post, because there are enough of them to be worth their own piece. The short version: do not ask on load, do not send daily, do not regenerate your VAPID keys, and do not treat your subscriber list as a channel you own outright — you are one bad month of over-sending away from losing it.
Running push on WordPress? EasyPush does everything in this post — soft-ask prompts with scroll and exit triggers, welcome notifications and honest CTR reporting, with frequency caps in Pro, and your subscriber list in your own database. Free, unlimited subscribers.
Frequently asked questions
How many people subscribe to push notifications?
It varies a lot by site and by how you ask. The post gives realistic ranges, but the number that matters is your own. Measure it after a few weeks.
How often should a blog send push notifications?
Start with two or three a week and watch your click rate. If it holds, you have room. If it falls, you’re sending too often.
When should the opt-in prompt appear?
After the reader has got something from the page, for instance halfway down a post, rather than the moment it loads. A custom prompt before the browser’s own dialog protects you from permanent blocks.
Which click rate should I trust?
Click rate measured against notifications displayed, not against notifications sent. Sent includes devices that never showed the message.
Can push notifications hurt my blog?
Yes, if you over-send. People who dislike a site’s notifications block it, and a block is permanent for that browser.