Frequency Capping: Stop Burning a Campaign on Ten Readers
Your most loyal readers will absorb a campaign whole if you let them, and the advertiser pays for every repeat.
The short version
- A frequency cap is a limit on how many times one browser is shown the same advertisement in a set period, usually written as impressions per day or per week.
- Without a cap, the readers who visit most often absorb most of a campaign, so the advertiser pays again and again to reach the same small group.
- Three impressions per day is a sensible starting cap for a month long banner campaign, and one per day is right when the advertiser wants breadth rather than repetition.
- A cap cannot deliver more people than the site has, so divide the impressions sold by the number of different readers the slot reaches and compare that to the cap before promising delivery.
- Caps are counted against a browser rather than a person, so one reader on a phone and a laptop counts twice and a reader who clears cookies resets to zero.
A frequency cap is a limit on how many times one browser can be shown the same advertisement in a set period, usually written as a number of impressions per day or per week. Without one, the readers who visit most often absorb most of a campaign, so the advertiser pays again and again to reach the same small group.
This matters more on a small site than on a large one. A network spreads a campaign across millions of people and the repetition averages out. Your site has a core of regulars who come back several times a week, and they will see everything you run.
What happens to a campaign with no cap
Traffic is never spread evenly across people. A minority of readers account for a large share of page views on almost any site that has been going a while, which is normally the best thing about it and is a problem here.
Here is an example with round numbers. A slot serves 30,000 impressions in a month, seen by 20,000 different readers. Those readers are not equal. A few hundred visit most days, a few thousand visit weekly, and the rest turn up once and leave.
One slot, one month, 20,000 different readers. An example with round numbers, not a measurement.
The advertiser bought 30,000 impressions, reached most of the audience once, and showed the same banner to 500 people roughly twice a day for a month.
Nobody wins in that picture. The advertiser paid for a month of diminishing returns on 500 people. Those 500 are your best readers, and they spent four weeks looking at the same artwork.
What does a cap actually change?
A cap does not create new readers. It stops a campaign spending itself on the same ones and pushes delivery out to people who have not seen it yet. Reach goes up, impressions per person come down, and the campaign takes longer to deliver.
That last part is the one that catches publishers out. A cap is a brake as well as a fairness setting, and if you set it hard enough the campaign will not finish. One calculation tells you in advance, and it takes a minute.
Divide the impressions you sold by the number of different readers that slot reaches in the same period. If the answer is higher than your cap, the campaign cannot deliver in full. Sell 30,000 impressions on a slot that 8,000 people see in a month and you need an average of nearly four impressions from each of them, so a cap of one a day is comfortable and a cap of two a month is a refund conversation waiting to happen.
What numbers should you start from?
There is no correct cap, only a sensible first guess that you move after a week of data. These are the ones worth guessing.
| Campaign | Cap to start with | Why |
|---|---|---|
| Month long banner, brand awareness | 3 per day | Enough repetition to be remembered, not enough to be resented. This fits most direct deals and is the right default. |
| Direct response with an offer and a deadline | 5 per day | Repetition is the point when there is something to act on and a date it stops. |
| Single sponsor of a section or a newsletter | 1 to 2 per day | The value is association, not volume. One placement a day reads as a sponsorship rather than an ad buy. |
| A campaign that must reach as many different people as possible | 1 per day | The lowest setting worth using. Go below it and the campaign underdelivers, which is worse for the advertiser than a second impression was. |
| A slot you are struggling to fill | No cap | A cap on inventory nobody else is buying costs you delivery and buys you nothing. |
Set one, wait a week, then read reach rather than impressions. These are starting points, not findings.
What a cap costs you
Be clear about the trade. A cap turns impressions you had sold back into impressions you have not sold. The reader still arrives, the page still loads, and the campaign you were counting on to fill that slot declines the request.
That is fine as long as the position has somewhere to fall. Every slot should have a fallback, and a capped campaign that skips a reader simply falls through to whatever is next in line: another advertiser, your network code, or a house ad. The fallback executes in the page rather than in an iframe, which is why AdSense and similar fill normally underneath. If you have not set one up, read how passback works before you start capping anything, because a cap on a slot with no fallback is just a blank box.
How do you set one and what do you tell the advertiser?
About ten minutes, once per campaign
-
1
Start at three a day
On a month long banner run this is the right first guess almost every time. Set it when you build the campaign rather than after the advertiser complains, because the first week is the week that gets burned.
-
2
After a week, read reach rather than impressions
Impressions only tell you the campaign is running. The number of different readers it reached tells you whether the cap is doing anything. If reach has barely moved while impressions climb, the cap is too loose.
-
3
Loosen it if delivery falls behind
Compare impressions delivered against where they should be a quarter of the way through. Behind by more than about 10 per cent means the cap is holding the campaign back, and one extra impression a day usually fixes it.
-
4
Put the cap in writing
One line on the insertion order: three impressions per browser per day. It takes a sentence and it stops the argument in month two about why the advertiser did not see their own ad more often.
Telling an advertiser the cap without being asked is worth more than it costs you. Most buyers on small sites have never been offered one, and a publisher who volunteers that they are protecting the audience from repetition sounds like somebody who plans to still be selling this slot next year.
The part that is actually a sales argument
Frequency capping sits with the rest of the targeting set: country, region and city, device type, device brand and model, ISP and carrier, day-parting and contextual keywords. It is included on both Pyrobid plans rather than sold as an upgrade, which matters, because the usual industry pattern is to charge for targeting once a publisher depends on it.
The reason to use it is not politeness. A campaign that reaches 6,000 different people once is a better result for the advertiser than one that reaches 900 people twenty times, and the advertiser who gets the first result is the one who renews. A cap is the cheapest renewal insurance you can set, and it takes about ten seconds.
Keep reading
Your own ad sales portal, live in five minutes
Advertisers sign up, add funds and build their own campaign. You approve the creative before it runs. Your existing ads keep running underneath as the fallback, so nothing comes down while you try it.
Start free trial14 days free. Your existing ads keep running. Cancel anytime.