Remarketing windows and frequency caps

Remarketing looks efficient because the numbers look good. People who already know you convert at a higher rate, so the cost per acquisition reads well.

A lot of that is people who would have come back anyway. The question worth answering is not what remarketing costs, it is what it adds.

Set the window to the buying cycle

The default window is almost always wrong because it is the platform's default, not your sales cycle.

A considered purchase with a three month decision process needs a long window, and a short one switches the ads off right when the decision is being made. A fast transaction needs a short window, and a long one spends money on people who decided weeks ago.

The way to set it is from your own data. Look at the time from first visit to enquiry across your last hundred leads. Most of them fall inside a range, and the window goes just past the far end of it.

Beyond that point you are paying to advertise to people who are not in the market.

Segment by what they actually did

One audience of all site visitors treats a bounce and a half-completed quote request the same way, which wastes most of the money.

Viewed a service page. Interested, early. Light frequency, broad message.

Spent real time on a page, or returned. Genuinely considering. This is the segment worth most of the spend.

Started a form or a quote and did not finish. The highest intent audience you have. Deserves its own creative and a message about the specific thing they abandoned.

Already enquired. Should be excluded from acquisition remarketing entirely and handled by sales. Continuing to advertise to somebody who is mid-conversation with your team is the most common waste in the channel.

Already bought. Either excluded, or in a completely separate campaign for the next purchase.

Each of those segments has a different window and a different cap, and running them as one audience averages all of it into something mediocre.

Cap the frequency before it damages the brand

Past a handful of exposures in a week, remarketing stops persuading and starts irritating. The person has registered the message, decided, and is now being followed.

The damage does not show up in the campaign numbers. It shows up in brand sentiment, in the people who hide the ads, and in the audience that stops responding to anything you run.

So there is a cap, and it is lower than feels comfortable. A few impressions a week on the considering segment, fewer on the early one. The abandoned-form segment can carry more because the message is still relevant.

Change the message as the window extends

The same ad for ninety days is the thing people notice and resent.

A sequence works better: the first week answers the obvious objection, the second shows proof, the third makes the offer concrete. After that, back off to a low frequency presence rather than continuing to push.

Which means remarketing is a small content programme rather than one ad with a budget.

Measure whether it is incremental

The only honest way to know what remarketing contributes is to hold out a portion of the audience and compare.

Suppress a share of the eligible audience for a period long enough to cover the sales cycle, then compare conversion rates between the two groups. If the difference is small, the channel is taking credit for conversions that were going to happen.

That test is uncomfortable because remarketing always looks good on the surface, and it is the only number in the channel worth trusting. Run it once a year, and size the spend to what it proves rather than to what the dashboard reports.