Ad Refresh: When It Adds Revenue and When It Hurts
What Ad Refresh Actually Is
Ad refresh means loading a new ad into a slot that's already on the page, without the reader loading a new page. The slot stays where it is; the creative inside it changes, and a new auction runs for the new impression. One slot, one pageview, several impressions.
That's why it's tempting. If a reader spends four minutes on a long recipe, a sticky sidebar unit that only ever shows one ad sits idle for most of that time. Refresh turns that time into more impressions.
It's also why buyers are wary of it. More impressions from the same reader doesn't mean more attention, and advertisers price refreshed impressions accordingly. Done carelessly, refresh can lower your CPMs, hurt viewability, make your site look like a "made for advertising" site, and break Google's rules. Done carefully, it can add real revenue on pages where people genuinely stay a while.
The Three Types of Refresh
Google Ad Manager's help article on declaring inventory that refreshes splits refresh into three types. It's a useful vocabulary even if you never touch Ad Manager, because buyers think in these terms.
- User action. The reader does something that brings up new content, and the ads change with it. Google's examples include filtering a list of products or picking a new radio station. Infinite scroll can count when a new article loads as the reader reaches the end of the previous one. What doesn't count: the reader simply being active on the page, or scrolling that doesn't surface substantially new content.
- Event-driven. The publisher changes the content, and the ads refresh with it. Think of a live sports score updating or the next video starting in an auto-playing playlist. The reader didn't ask for it, but the page genuinely changed.
- Time-based. The ad changes after a set interval, regardless of what's happening on the page. This is what most people mean when they say "ad refresh", and it's the one that gets publishers in trouble.
You'll also hear about viewability-based or scroll-based refresh. That's not a fourth trigger but a time-based timer that only runs while the ad is on screen. For Google's declarations it's still time-based, just a far more defensible version.
Refresh types compared
| Type | Trigger | Ad Manager web minimum | Buyer perception |
|---|---|---|---|
| User action | Reader loads new content (filter, next article) | None | Closest to a fresh pageview |
| Event-driven | Publisher updates content (live score, next clip) | 30 seconds | Acceptable if the page truly changed |
| Time-based | A fixed timer | 30 seconds | Most likely to be discounted |
| Time-based, viewability-gated | Timer that runs only while the ad is in view | 30 seconds (declared as time-based) | Better than a blind timer |
What Google Says About Refresh
Google's rules depend on which Google product is serving the ads, and the difference is big.
AdSense: you don't refresh it yourself
The AdSense ad placement policies are blunt: publishers may not refresh a page or an element of a page unless the user requests a refresh. That covers placing ads on pages or in placements that auto-redirect or auto-refresh. In practice, if you've pasted AdSense code onto your site, don't wrap it in a script that reloads the unit every 30 seconds, and don't put it on a page that reloads itself on a timer. That's a policy violation, and it's the kind that can put an account at risk.
Google doesn't publish a detailed list of which interactions count as a user request for AdSense, so the safe choice is not to build refresh logic around AdSense units at all.
Ad Manager: refresh is allowed, but you must declare it
Google Ad Manager does allow refresh, with conditions. Its help article says you must set up inventory rules that declare which ad units refresh, what triggers the refresh, and the minimum interval. Those declarations live under Inventory, then Inventory rules, then Publisher declarations. They're passed to buyers through Google's real-time bidding protocol so they know they're bidding on refreshing inventory.
A few details from that article worth knowing:
- Failing to declare refresh, or declaring it wrongly (for example the wrong trigger type or interval), is described as a violation of Google policy.
- The declaration doesn't make anything refresh. It's a label for buyers. Your actual refresh logic still lives in your ad code or your wrapper.
- Minimums on desktop and mobile web: no minimum for user-action refresh, 30 seconds for event-driven, and 30 seconds for time-based.
- Buyers can and do respond to the label. Google says demand restrictions apply to every request that matches your declaration's targeting. If you declare refresh across your whole network, even units that never refresh get treated as refreshing. Target the declaration at the specific ad units or key-values that actually refresh.
Managed networks and wrappers
If you're with a managed partner such as Mediavine, Raptive, Ezoic or Freestar, refresh is usually part of their setup, not yours. Don't layer your own refresh script on top of theirs. Ask them which units refresh, on what trigger, and whether you can turn it down or off. The network's own account team is the authority on how their refresh works.
Why Refresh Can Lower What You Earn
Buyers discount refreshed impressions
The first impression in a slot arrives when the reader has just landed and is looking at the page. The fifth impression in the same slot arrives after the reader has been scrolling past it for minutes. Buyers know the difference. Some exclude declared refresh inventory from their campaigns entirely; others bid lower on it. How much lower varies by buyer and by site, and no one publishes a reliable industry-wide number, so be skeptical of anyone quoting one.
So three refreshes don't mean four times the money. They mean one full-price impression plus some cheaper ones, and if your setup also drags down the price of the first impression (by mis-targeting the declaration, for instance), you can end up with more impressions and less money.
Viewability can go up or down
The common industry standard for a viewable display impression, set by the Media Rating Council, is at least 50% of the ad's pixels on screen for at least one continuous second. Viewability rate is viewable impressions divided by measured impressions.
Refresh can push that rate either way. A blind timer that refreshes units the reader scrolled past minutes ago creates lots of impressions no one could see, so your viewability rate drops, and buyers who bid on viewability pay less across the board. A refresh that only fires while the ad is on screen does the opposite: the new impression starts life in view, so it's very likely to count as viewable. That's the single biggest argument for gating refresh on viewability.
Refresh is one of the "made for advertising" signals
In 2023 a group of advertiser trade bodies led by the ANA published a definition of made-for-advertising (MFA) sites: sites built mainly to arbitrage ad inventory. One of the listed characteristics is rapidly auto-refreshing ad placements. The others include a high ad-to-content ratio, a high share of paid traffic, generic content, and templated design.
The definition isn't a checklist where one trait condemns you, and the groups behind it didn't say how many traits a site needs to show. But buyers and verification vendors use MFA lists to block inventory, and aggressive refresh on a page with thin content and heavy ad density is exactly the combination that gets a site flagged. If you're still building up a site, your bigger lever is content; see content quality for ad approval.
It costs page performance
Every refresh runs a new auction and renders a new creative. On a page that's already struggling, that shows up as worse responsiveness. See why site speed tanks after adding ads before adding refresh to the mix.
Conditions a Responsible Refresh Setup Checks
None of these come from a single official rulebook (apart from the 30-second minimum in Ad Manager). They're widely used practice, and they all point the same way: only refresh when a real person is plausibly looking at the ad.
- The ad is in view. Use the same threshold buyers use: at least half the ad on screen. If you use Google Publisher Tag, the
slotVisibilityChangedandimpressionViewableevents tell you when a slot is in view and when its impression became viewable. - The tab is visible. If the reader switched tabs, the ad isn't being seen, even if it's technically "in the viewport". The browser's Page Visibility API (
document.visibilityState) tells you this. Pause the timer when the page is hidden. - Enough on-screen time has passed. Count only the time the ad has actually been in view on a visible tab, and never go below 30 seconds if you're in Ad Manager. Many setups use longer intervals; slower refresh means fewer impressions, but each one is worth more.
- A cap per slot and per page. Stop refreshing after a fixed number of cycles. A tab left open overnight shouldn't generate hundreds of impressions.
- The right slots. Sticky sidebar units and anchor units that stay on screen are natural candidates. In-content units that scroll past quickly rarely build up enough on-screen time.
Should this slot refresh now?
- Is the ad in view?At least 50% of its pixels on screen. If not, pause the timer.
- Is the tab visible?Check the Page Visibility API. If hidden, pause the timer.
- Has enough in-view time passed?30 seconds minimum in Ad Manager; only count visible, in-view time.
- Has the cap been reached?If this slot or page has hit its limit, stop for good.
- Refresh the slotRun a new auction, reset the timer, add one to the count.
How to Test Whether Refresh Earns You Anything
Refresh changes how many impressions a pageview generates, so you can't judge it by impression counts or CPM alone. More impressions at lower CPM can mean more money or less. Here's a clean way to find out.
Split your traffic
Turn refresh on for a portion of your traffic, not all of it. Assign the split per visitor or per session, not per pageview, so the same reader isn't flipping between versions. A simple way is to set a key-value (for example refresh=on or refresh=off) on each ad request so you can separate the groups in your reporting. Run both groups at the same time, so seasonal swings hit both equally.
Watch session RPM, not page RPM
Page RPM is revenue per thousand pageviews. Session RPM is revenue per thousand sessions. Refresh can raise page RPM while lowering the thing you actually care about. If refresh makes the page heavier or more annoying, readers may view fewer pages per visit. Page RPM goes up because each page earns more, while session revenue goes down because each visitor sees fewer pages. Session RPM catches that. Our guide to maximizing ad RPM explains the difference in more detail.
Check the side effects too
- Viewability rate for the refreshing slots, before and after.
- CPM on the first impression of each slot. If it drops in the test group, your declaration or buyers' reaction is costing you on impressions you were already getting.
- Pages per session and engagement time. If they fall, readers are noticing.
- Core Web Vitals from real-user data, especially responsiveness.
Give it time
Run the test over full weeks, since weekday and weekend traffic often differ, and long enough for bidding to settle after a new declaration. A small site needs longer to get numbers that aren't noise. If the session RPM gap between groups is small and flips from week to week, you don't have a result yet.
A hypothetical example: a recipe site turns on viewability-gated refresh for its sticky sidebar in half its sessions. In the test group, page RPM rises and the sidebar's average CPM drops. If pages per session hold steady and session RPM ends up higher, refresh stays on. If pages per session fall enough to cancel the gain, it goes off, even though page RPM looked better.
When Refresh Is and Isn't Worth It
Refresh tends to make sense when readers genuinely stay on one page for a long time (long guides, recipes, live updates, tools people keep open), you have a slot that stays on screen while they do, you're on Ad Manager or a managed partner that handles declarations properly, and your content would never be mistaken for an MFA site.
It tends not to be worth it when your pages are short and readers leave in under a minute (there's nothing to refresh), you're on AdSense alone (you're not allowed to), your site is new and still being reviewed by networks, or your ad density is already high.
If you're earlier in the journey and still working toward approval, refresh isn't your problem yet. Run your site through AdGateScore's checker to see which networks you're ready for, and use the revenue calculator to get a feel for what better RPM or more traffic would actually be worth before you spend time tuning refresh intervals.
We build AdGateScore, the ad network readiness scanner. Our articles draw on the requirements each network publishes and on what our scans find.