Ad refresh and CPM: does refreshing ads actually make more money?

Ad refresh is requesting a new ad for a slot that is already on the page, so one page view produces several impressions instead of one. On content where readers stay a long time, guides, wikis, live blogs, forums, it is the difference between monetizing a visit once and monetizing it repeatedly.

The mechanism is simple and the discipline is not. Every extra impression divides the same attention, so a refresh strategy that ignores whether anyone is looking manufactures inventory nobody sees, and buyers price that in.

Both, and which one wins depends on the trigger. Refreshing raises your total impressions, so revenue can go up even as CPM goes down: what matters is revenue per session, not the headline CPM.

The risk is that a poorly triggered refresh lowers your viewable rate, and viewability is what buyers increasingly bid against. Manufacturing unviewable impressions therefore hits the price of every impression you sell, including the good ones.

The constraints most demand partners expect, and the reasons behind them:

• Refresh only when the slot is in view. This is the rule everything else rests on.
• Leave at least thirty seconds between refreshes. Faster tends to breach partner policies and buys you complaints instead of money.
• Pause when the tab is hidden. A background tab is not attention.
• Cap the number of refreshes per slot per session, so a long visit does not turn into an impression farm.
• Do not refresh formats sold on completion or on engagement unless the partner allows it.

Check your own demand partners' policies before deploying: several allow refresh only under explicit conditions, and violating them can cost you the demand rather than the CPM.

On short visits, first, and this is arithmetic rather than strategy: a reader who leaves in twenty seconds never reaches the second impression, so refresh adds nothing to news-style traffic that bounces.

And on blocked traffic, which is the part publishers miss. If a visitor's blocker stops your ad library from loading, there is no slot, no first impression, and therefore nothing to refresh. Refresh multiplies inventory you have; it cannot create inventory that never existed.

That distinction matters when you model the upside. Refresh applies only to the share of your traffic that already sees ads. If a fifth of your page views never make an ad request, your refresh gains apply to four fifths of your audience, and the other fifth needs a different fix entirely.

Judge it on revenue per session, not on impressions or CPM in isolation. Impressions will rise by construction and CPM will usually dip; neither number answers the question on its own.

1. Baseline revenue per session for a full week before enabling anything.
2. Enable refresh with an in-view trigger on one template or one section, not site-wide.
3. Compare revenue per session, viewable rate and engagement against the baseline.
4. Watch bounce and time on page: if refresh coincides with readers leaving sooner, you bought impressions with attention.

Run it a full week either way. Ad blocking, and therefore your real impression base, varies by day of week and by traffic source, so a three-day test measures the calendar as much as the change.

Refresh is optimization of inventory you already have. It does nothing for the page views that never entered the auction, and on many publisher audiences that is the larger number.

The order that makes sense is therefore: measure how much of your traffic is blocked, recover what is recoverable, and then optimize the resulting inventory with refresh and placement. Doing it the other way round means tuning the visible slice while the invisible slice keeps growing.

If you have never measured the blocked share, that is the cheap first step: our script reports detections, recoveries and recovered revenue per site from a one-line install, and what adblock costs publishers shows how to turn that into a number you can plan with.