The two-week ad slump, and how to tell fatigue from a bad ad.

Somewhere in the second or third week, the numbers start to slide. The ad that opened strong, clicks coming in cheap, conversions steady, begins costing more for the same result. A week later it's clearly worse than it was. The instinct is to blame the ad: maybe the offer was weaker than you thought, maybe the image doesn't land the way you hoped. But before you touch the creative, the real question is: has this ad actually failed, or has it just been seen too many times by the same people?

Those are different problems with different fixes, and a dashboard showing declining clicks or conversions can't tell you which one you're looking at. You have to go looking for the difference yourself.

What fatigue actually looks like inside the numbers.

Meta's research team studied this directly, and the scale of repeat viewing surprised even them. Across their ad delivery data, when an ad is shown, the person seeing it has, on average, already seen that exact same ad 4.2 times before. More strikingly, over 19% of all impressions are someone's fifth time or later seeing that exact ad. That's not a fringe case. That's roughly one in five views happening after the ad has already made its case four times and failed to convert.

Meta's team defines creative fatigue narrowly and usefully: it's the decline in how well an ad performs for the same individual person as their exposure count climbs, separate from anything happening at the audience level. That distinction matters, because a dashboard blends both together. Overall performance can drop because you're running out of fresh people to show the ad to, or because the people you're still showing it to are tired of seeing it. Only one of those is about the ad. Figuring out which cause is dragging your numbers down takes knowing your business and your audience, not just the ad, which is why Knowing learns your business first and judges the ad second.

The test that separates fatigue from a real miss.

Meta ran a controlled experiment: take ad sets that had accumulated high fatigue, inject brand-new creative, and measure what happened. The fix worked in proportion to how tired the ad set actually was. The higher the fatigue level going in, the bigger the lift after refresh. Across the high-fatigue cases, swapping in new creative improved conversion rate by an average of 8%.

That dose-response pattern is the tell. If a fresh creative swap produces a big jump specifically where fatigue was worst, the decline was caused by repetition, not by some deeper flaw the audience finally noticed. Meta's guidance to advertisers follows directly from this: the fix isn't tweaking the existing ad's headline or color, it's introducing genuinely new creative of equal or greater quality, or moving to a new audience that hasn't already been shown the old one four times.

A dashboard can't tell you whether an ad failed or whether it was just seen too many times. You have to go looking for the difference yourself.

Not every viewer just gets bored. Some get annoyed.

The word 'fatigue' suggests something gentle, like a viewer's attention slowly running out of gas. Here's where it gets sharper: a 2018 study in the Journal of Marketing Research, by Inyoung Chae, Hernán Bruno, and Fred Feinberg, found something hiding inside the average that a single blended number can't show. Using a dataset covering more than 12,000 users across over 400 websites, they built a model that sorts viewers into distinct response groups rather than averaging everyone together.

Most groups showed ordinary wearout: each additional exposure did a little less than the one before, eventually flattening to near zero. But one group, accounting for roughly 24% of the sample, showed something the researchers call weariness: additional exposures made them less likely to visit the site, a genuinely negative effect, not just a diminished one. In this dataset, roughly a quarter of viewers seeing an ad repeatedly weren't tuning it out. They were pushing back against it.

The researchers were explicit that this finding only shows up if you account for the fact that different people respond differently; a model that pools everyone into one average erases it completely. That's a useful warning for anyone reading their own ad metrics as one blended number. The same weekly average can hide that same roughly 24% of your audience actively souring on the ad while the rest stay neutral.

The same study also surfaced a related but separate finding, worth knowing before you decide where to run the refreshed ad: publisher quality varies enormously. The single best-performing publisher in their data was nine times more effective than the one ranked 26th. And a frequency cap tuned to how different viewers actually respond, rather than one blanket limit for everyone, improved overall results by as much as 15%. Fatigue isn't only about the creative. It's also about how often, and where, that creative keeps showing up to the same eyes.

Some ads were never built to survive repetition.

Here's where it gets more useful for judging your own creative, not just diagnosing the symptom. A 2013 study in the International Journal of Advertising, by Kevin Lehnert, Brian Till, and Brad Carlson, tested real TV commercials embedded in actual program content, then followed up a week later. They compared highly creative ads against more conventional control ads across repeated exposure.

The obvious finding: creative ads were recalled better than the control ads, at least at first. The less obvious finding is the one that matters here. That recall advantage shrank as exposures repeated. Creativity is not a permanent shield against fatigue. It buys you a bigger lead early on, and that lead erodes the same way everything else does.

But there's a second layer, and it's the one worth sitting with. Creative ads climbed to peak effectiveness quicker than conventional ads, what the researchers call faster wearin. And they were also less susceptible to wearout over repeated viewing. In their words, creative ads "demonstrate wear in effects more quickly and are less susceptible to wearout." Creativity doesn't just affect where an ad starts. It changes the shape of the whole curve, front-loading its impact and stretching out its useful life.

What this means for the ad sitting in front of you right now.

If your ad performed well for a stretch and then faded fast, and the fade lines up with your frequency numbers climbing past four or five views per person, that's the fatigue pattern Meta's data describes. The fix is a genuine creative swap or a new audience, not a tweak to the one that's already worn thin.

But if the ad faded slowly from a modest peak, never climbed much to begin with, or fatigue is low and it's still underperforming, you're probably looking at something closer to the weariness Chae, Bruno, and Feinberg found: a real, negative reaction to the ad itself, not exhaustion from overuse. That's not solved by frequency caps or fresh audiences. That's a creative that never worked, not one that wore out.

The shape of the curve is the evidence. A steep climb followed by a steep fade, right where exposure counts are highest, points to fatigue. A flat or shallow curve that never really worked points to a creative that needed rebuilding before it ever ran, not a refresh after the fact.