When to Stop Scaling a Winning Ad

The hard part of a winner is not finding it. It is admitting when it is finished. Ads that carried an account for two months have a way of getting the benefit of the doubt for a third, and the third month is usually where the money goes.

Scale in steps you can undo

We raise budget on a winning ad by around 20% every second day and watch the two days after each raise, not the two hours. Doubling a budget overnight throws the ad back into learning and costs you the result you were trying to buy more of.

The three readings that say stop

Frequency past three on the prospecting audience with no new audience to move into. Cost per acquisition up by a third from its best week and still climbing after a raise is reversed. Hold rate at three seconds falling while the click-through rate stays flat, which usually means the people left have already seen it.

Retire it properly

A finished ad is not a dead ad. Pull it out of prospecting, keep it running to your warm audiences at a small budget, and put the concept back on the shoot list with one thing changed. Half the winners we have made this year were the second version of something that had already stopped working.

Have the next one ready

This is the whole argument for a steady production schedule rather than a campaign every quarter. If you only shoot when performance drops, you are always six weeks behind the account. If you shoot every month, retiring a winner is a decision, not an emergency.

We tell clients to expect a good ad to last four to eight weeks in prospecting. Anything longer is a gift, not a plan.