How to tell creative fatigue from other causes of falling ROAS, estimate how fast your ads wear out at your budget, and keep fresh creative coming before results drop.
Picture a campaign that launches with a return on ad spend (ROAS) of 4, meaning €4 in sales for every €1 of advertising. Six months later it sits at 1.8, and the usual suspects get the blame: rising ad prices, an algorithm update, a slow season. Often the real cause is simpler. The same three ads have been shown to the same people for months, and those people stopped noticing them long ago.
That is creative fatigue: people see the same ad so often that they stop responding, so clicks fall, each sale costs more and ROAS slides. You fix it by measuring how quickly your ads wear out, retiring them on a clear rule rather than a hunch, and producing new ones on a steady schedule so a replacement is always ready. Better ads help, but a reliable supply of new ones matters more.
Fatigue costs you twice: you pay more for every sale, and you miss the sales a fresh ad would have won. This article shows you how to separate fatigue from other causes of falling returns, how to estimate how fast your ads will tire at your budget, how to split spend between proven and new creative, and how to set up a production rhythm that keeps pace.
Before you order new ads, check that fatigue is really the problem. Returns also fall when auction prices rise, when your website converts worse, when a bestseller goes out of stock or when tracking breaks, and new creative fixes none of those. Look at four numbers together: frequency (the average number of times each person has seen an ad), click-through rate (CTR, the share of viewers who click), cost per thousand impressions (CPM) and the conversion rate on your site.
A practical rule is to compare each ad with its own first fortnight, not with the account average. If its frequency has risen, its CTR has dropped by about a third from that early level and its cost per sale has climbed, treat it as worn out and schedule a replacement. Adjust the threshold once you have a few months of your own data.
How quickly an ad tires depends on how often each person sees it, and you can estimate that in advance. Divide your monthly spend by the CPM and multiply by 1,000 to get your impressions. Divide the impressions by the size of the audience, and you have the average number of times each person sees your ads in a month.
Here is a worked example with round figures. You spend €5,000 a month on remarketing, showing ads to 50,000 recent website visitors at a CPM of €10, which buys 500,000 impressions: about 10 per person per month, or more than two a week. Put the same €5,000 into a broad audience of two million people at an €8 CPM and you get 625,000 impressions, roughly one for every three people each month. The first audience is likely to tire of a single ad within a few weeks, while the second can see the same ad for months.
Treat these figures as a minimum. Platforms do not spread impressions evenly; they concentrate them on the people most likely to buy, so your best prospects see an ad more often than the average suggests. Check frequency over the last seven days rather than over the ad's lifetime, because a lifetime figure hides a recent surge.
Smaller markets tire faster. If you sell in Austria, Slovenia or the Baltic states, or split Belgium or Switzerland by language, each audience is a fraction of what a campaign in Germany or France would reach. The same budget per market then means a much higher frequency, so either refresh creative more often in small markets or give them smaller budgets.
Once you spend more than about €20,000 a month, a simple rule of thumb keeps performance and renewal in balance: 70% of spend on proven ads, 20% on variations of proven ideas and 10% on genuinely new concepts. The exact numbers matter less than the principle. A fixed share of your budget always goes to finding the next winner, so you are never caught without one.
The 20% is where many of your future winners will come from. A variation keeps a message that works but changes what people notice first: a new opening line, a different person on screen, a carousel instead of a video, or a vertical edit for stories and short-form video. The 10% tries different angles altogether, such as a comparison with the alternative, a founder explaining why the product exists or a message written for one customer group.
On €20,000 a month, the split is €14,000, €4,000 and €2,000. If a sale costs you around €40, the €2,000 test budget buys about 50 sales a month, which is enough for a rough read on two or three new concepts, not ten. With a smaller budget, test fewer things at once and judge them on early signals that arrive in larger numbers, such as CTR or cost per add-to-basket, then confirm the winners on sales.
Winners move up. A variation that beats its parent takes over part of the 70%, and the ad it replaces is retired, which keeps the proven group from slowly going stale.
People do not notice small changes. A different button colour or a reworded last line looks like the same ad, so it does little to reset fatigue. Ad platforms also look for varied ads to show to different people, and a set of near-identical versions gives them very little to choose from.
A real refresh changes what someone sees in the first two or three seconds: the opening image or line, the person, the setting, the format or the angle. A simple check is to put the old and new ads side by side as small thumbnails. If a stranger could not tell them apart at a glance, your audience will not either.
This also affects what you get for your production budget. One shoot can yield many genuinely different ads if you plan the openings and angles before filming, while ten colour variants of one design are, in practice, a single ad.
Teams that treat creative as a quarterly project tend to run out of fresh ads halfway through the quarter and then rush. The alternative is a weekly rhythm sized to your own numbers. Count how many ads you keep live, find out how many weeks your ads typically last before they hit your fatigue rule, and divide the first number by the second.
Say you keep eight ads running and they last about four weeks each. You need two replacements a week just to stand still, plus one or two new concepts for testing, so three or four new ads a week in total. That is a plan you can staff and budget for, and it rests on four parts.
The economics usually justify the effort. Suppose you spend €10,000 a month and fresh creative keeps your cost per sale at €40 instead of letting it drift to €60. That is 250 sales a month instead of about 167, roughly 80 extra sales from the same budget, which pays for a good deal of editing time.
Over time the log shows which angles last longest and which formats your audience tires of quickly, so each brief gets sharper. It also changes how you see your top ad.
Your best-performing ad is rarely your best idea; it is simply the one your audience has not tired of yet.