Social isn't suffering from creative wear out - it's suffering from audience saturation

By Jo-Ann Foo, Senior Director, Analytic Partners 

Your ad worked brilliantly last week. This week, the numbers are slipping - and it's not looking good for next week. The usual diagnosis? Creative fatigue - the audience has seen the ad too many times, the idea has worn out, and it's time for something new.

Sometimes, that diagnosis is right. But increasingly, marketers may be mistaking one problem for another. The creative itself hasn't necessarily worn out, but the platform you're using may have narrowed delivery to the point where the same audience has been saturated.

And for the platforms that now account for such a significant part of most media plans, that distinction matters.

Consider this scenario. You're working at a shaving brand that needs to reach roughly 270 million adults who might need some kind of shaving product. You set the campaign up for conversions and share its e-commerce data with the platform so the algorithm can find people who look like existing customers.

Within days, that alone collapses the audience from 270 million down to around 50 million. The platform keeps going from there, favouring whoever watches, clicks or converts first. Delivery is tightened around an ever-smaller group with each passing day. The frequency climbs, costs rise, and engagement softens.

On the surface, that looks exactly like creative fatigue. It isn't. The platform has simply exhausted the easiest audience for that one version of the idea, and kept hammering it.

What I'd call that is audience saturation.

None of this is unique to one campaign. Unlike TV, where the same broad audience keeps watching no matter how long a campaign runs, social's delivery narrows from the moment a campaign goes live.

TV can run one ad broadly for years and still reach everyone it needs to. Social can't - it needs more executions to do the same job, and each one still has to answer to the same idea, or the brand fragments right when the algorithm needs it to hold together.

It doesn't matter what you're optimising for, and it doesn't matter if your KPI is brand or sales. The algorithm will fold in on itself chasing whichever signal you've given it. Conversion data will sharpen the signal, while a brand campaign narrows it more slowly - but it will narrow all the same.

The data backs it up

This isn't a one-off pattern. Analytic Partners analysed 51,218 pieces of creative and found genuine wear-out in just 14 of them, a finding System1 has landed on independently.

We can already see proof of this. Coca-Cola's "Holidays Are Coming" campaign has run largely unchanged since 1995 and still scores System1's maximum rating for effectiveness nearly three decades later.

More locally, we're seeing companies catching on - just last year, Woolworths decided to rerun 2024's Christmas ad, redirecting the funds it would have used on a new ad to measures dedicated to cost-of-living relief.

Creative doesn't decay anywhere near as often as marketers assume. What decays is the size of the audience still seeing it.

The fix is recasting, not reshooting

The answer isn't to abandon the campaign. It's to give the algorithm more entry points into the same idea, like different ages, ethnicities or variations of facial hair like beards, moustaches, sideburns, trimmed down to clean-shaven.

It's the same product with the same tone and brand codes throughout but each new cut hands the algorithm a different signal to chase. So instead of tightening further around the same exhausted pool, it starts responding to people who look like the new casting, and delivery opens back out.

You're not reshooting a new ad. You're just giving the algorithm more people who look like they matter, so it stops collapsing onto the ones it's already spammed to death.

Doing this well means integrating it into planning and production from the start and running a shoot day that captures five or six variations rather than one, rather than bolting it on after performance dips and the budget's already been spent.

Meta's own analysis backs up this dynamic. The average Meta impression is served to someone who has already seen that creative 4.2 times in the previous 30 days, and more than 19% of impressions go to people who have seen it more than five times

In high-fatigue cases, adding new creative variations lifted conversion rate by an average of 8%. That doesn't mean killing the campaign idea; it simply means giving the algorithm more versions of it to work with.

Generative AI makes this easier to do, and easier to get wrong. It's the same shift Coca-Cola made with its own AI-rebuilt Christmas ad - the tool lowers the cost of producing new versions, so a brand can either scale disciplined variations or flood the platform with disposable ones.

The opportunity isn't endless different ads but AI-enabled variation that still answers to the same distinctive brand idea.

This isn't to say the algorithm is the villain here. It's built to find efficiency, not audience health, and those are two different things.

So when performance starts to decline, marketers need to look beyond the creative itself and understand just how far delivery has narrowed and how heavily the remaining audience is being saturated. By getting that diagnosis right, you'll be solving the right problem and often with a considerably cheaper fix.