Ad refresh means reloading the creative in a slot while the reader stays on the page — every 30 seconds, say, or when they scroll back to it. It doubles or triples your impression count on long-session pages, which is exactly why it needs thinking about carefully.
Why anyone does it
If a reader spends four minutes on an article, a static slot serves one impression. The reader had four minutes of exposure; the sponsor paid for one impression. Refresh recovers some of that.
On sites with genuinely long sessions — documentation, forums, live scores, anything with an embedded video or a tool people use — refresh can be the difference between an ad slot being worth selling and not.
Where the line is
The honest version of refresh has all of these:
The slot is actually in view. Refreshing an ad nobody is looking at manufactures impressions from nothing. Use an IntersectionObserver and only refresh visible slots.
The tab is active. A backgrounded tab should not be counting anything. Check document.visibilityState.
The interval is long enough to be seen. 30 seconds minimum. Anything under that is not exposure, it is counting.
There is a cap. Stop after four or five refreshes. Someone who has been on the page ten minutes has stopped looking at your sidebar.
The reader is present. If there has been no scroll, mouse movement or keypress in a couple of minutes, pause.
Miss any of those and you are inflating impressions, which is fraud whether or not you thought of it that way.
Tell your sponsors
This matters more than the technical details. If a sponsor buys 50,000 impressions and 20,000 of them are refreshes to the same people, they are getting far less reach than the number implies.
Put it on the rate card:
Slots refresh every 30 seconds while in view, capped at 4 refreshes. Roughly 1.8 impressions per pageview.
A sponsor told this upfront will price accordingly and trust the rest of your numbers. A sponsor who works it out later from a suspicious frequency figure will not buy again.
When not to refresh
Short-session sites. If your average time on page is 90 seconds, refresh gains you almost nothing and costs you credibility.
Sponsorship deals sold by time, not volume. If someone bought "the sidebar for March", refresh changes nothing for them and only pads a number.
Anywhere you sell on click rate. Refresh reliably drives click rate down, because the denominator grows and the numerator does not. A slot with a 0.6% CTR can look like 0.25% purely from refresh, and that is the number a sponsor will judge you on.
That last one is the real trap. Refresh raises impressions and lowers every quality metric derived from them. If your sponsors care about clicks — and direct sponsors usually do — refresh can make a well-performing slot look bad.
The recommendation
For most small publishers: don't. The impression gain is modest, the CTR damage is real, and the disclosure burden is a permanent tax on every conversation.
Refresh earns its place on sites where people genuinely stay for many minutes on one page. Everywhere else, a second well-placed slot is worth more than refreshing the first one.