Your Best Ad Died Two Weeks Ago. You Just Haven't Noticed Yet.

You found a winning ad. The clicks were cheap, the leads were flowing, the cost per acquisition was right where you wanted it. So you did what any sensible business owner would do: you let it run.

That was three months ago. And right now, that ad is costing you more than the day you launched it, delivering fewer leads, and training the algorithm to think your business has nothing new to say. You just can't see it yet because the decline is gradual enough to mistake for "the market getting more competitive."

It's not the market. It's the ad. And the data says it probably stopped working two weeks after you launched it.

The Shelf Life Problem

The same winning creative that might have scaled for 20-25 days a year ago now maxes out in 7 to 12 days. If you're reaching 50% or more of your target audience weekly, fatigue arrives within a week. Nielsen's 2025 Digital Ad Effectiveness report found that ads lose impact up to 35% sooner in algorithm-driven campaigns than in manually structured ones.

Here's the uncomfortable part: fatigue no longer announces itself cleanly. It used to show up as rising frequency and falling click-through rates. Now, according to Meta's own updated advertiser guidance, performance degradation increasingly appears through declining conversion efficiency while surface metrics still look acceptable. Your CTR might hold steady while your cost per lead quietly doubles.

The clearest warning signs in 2026: a 20%+ week-over-week drop in CTR, rising CPMs without new competitors entering the auction, and ad frequency exceeding 3.0 on cold audiences. But by the time those signals are obvious, you've already been overpaying for weeks.

This is why so many businesses feel like their ads stopped working when really it's a single creative that expired.

Meta's Algorithm Now Punishes Sameness

There's a deeper problem beyond simple fatigue, and it has a name: Andromeda.

Andromeda is Meta's ad retrieval engine, rebuilt from the ground up using deep neural networks. It replaced the rule-based system that used to decide which ads got shown. The old system matched audiences to ads. Andromeda matches creatives to users, processing 10,000x more data per impression than its predecessor.

The critical concept for advertisers is the Entity ID. Every ad you create gets assigned a semantic fingerprint. Andromeda uses this fingerprint to determine whether your "new" ad is genuinely different from your existing ads or just a cosmetic variant.

If you upload five ads where you've changed the headline colour, swapped a single word, or adjusted the background music, Andromeda clusters them under a single Entity ID. As Jon Loomer explains, those five ads get one ticket to the auction. Not five.

A controlled test cited by Confect.io showed that one ad set with 25 diverse creatives beat a fragmented 5-ad-set structure by 17% on conversions and 16% on cost. Fewer ad sets, more genuinely diverse creative within them.

This is a fundamental shift. Creative has effectively replaced what audience targeting used to do. Your creative IS your targeting now.

Why the Science Says Variety Beats Optimisation

Here's where it gets interesting. The platform data aligns perfectly with what marketing science has been saying for decades.

Byron Sharp's research at the Ehrenberg-Bass Institute established that brands grow by being easy to think of in buying situations. He calls this mental availability. But mental availability isn't built by showing the same message to the same people repeatedly. It's built by linking your brand to multiple Category Entry Points, the different needs, occasions, and motivations that trigger someone to think about your category.

A plumber, for example, enters someone's mind through dozens of different doors: a dripping tap, a bathroom renovation, a burst pipe at 2am, a new house build, water pressure problems, a rental property inspection. Each of those moments requires a different message to be relevant. One ad, no matter how good, can only walk through one door.

Sharp's data across 130+ brands shows that brands with higher mental availability are linked to more buying situations, not linked more strongly to one. The implication for advertising: a portfolio of different creative angles, each addressing a different entry point, builds more mental availability than a single polished ad repeated endlessly.

This is why your best ad copy often comes from your customers' own language. Different customers describe their problems in completely different ways. Each description is a potential creative angle.

Rory Sutherland, Vice Chairman of Ogilvy, offers the strategic explanation for why this variety matters competitively. His eighth rule of alchemy: "Test counterintuitive things only because no one else will." Most businesses find a winning formula and optimise it to death. They tweak headlines, adjust colours, A/B test minor variations. Sutherland argues this is a trap: "It doesn't pay to be logical if everyone else is being logical." The competitive advantage lies in trying approaches that feel risky or unusual, precisely because competitors won't.

Applied to creative strategy, this means the most valuable ads in your portfolio aren't the safe iterations of your best performer. They're the weird ones. The testimonial video that breaks every "best practice." The static image that leads with a counterintuitive claim. The ad that talks about your category in a way nobody else does. Those are the ads that Andromeda treats as genuinely new Entity IDs, and the ones Sharp's research suggests will reach buying situations your current creative misses entirely.

What "Creative Diversity" Actually Means

Sam Tomlinson, EVP of Warschawski and one of the sharpest thinkers on creative performance, developed a framework that makes this practical. He calls it the 3-Lens Framework, and the formula is simple:

The most fundamental error, he argues, is confusing optimisation with innovation. Running A/B tests with barely perceptible modifications (headline variants, colour shifts, minor copy changes) is optimisation. Launching a completely different creative concept, a new format, a new angle, a new hook, is innovation. Andromeda rewards the latter.

Jon Loomer puts it bluntly: "Entity ID diversification is concept diversification. Not colour tweaks. Not headline swaps. Concept."

What This Means for Your Business

You're probably thinking: "We're spending $3,000 a month on ads. We can't produce 27 creative variants."

Fair. But you don't need 27 on day one. You need a system that replaces dying creative before it drags your results down. Here's what that looks like for an SME budget:

A different concept means a different hook, a different visual approach, and ideally a different format. A testimonial video, a problem/solution static, a before/after carousel, and an educational text-overlay ad are four different concepts. The same hero image with three different headlines is one concept.

This replaces the "set and forget" approach. You don't need a production studio. A phone-recorded customer testimonial, a simple static with a bold claim, and a screen recording of your Google reviews are all viable creative assets that cost nothing to produce. The bar for "production quality" in 2026 is lower than you think, as long as the concept is genuinely different.

If your cost per lead has climbed 20%+ while CTR holds steady, your creative is fatigued. Don't wait for the obvious signals. Check weekly.

Tomlinson's portfolio analogy is the right mental model. A good portfolio has diversified assets, not everything riding on one winner. Your best-performing ad should never represent more than 30-40% of your total spend. The rest should be spread across different concepts, formats, and angles.

We've written before about why single points of failure are so dangerous in marketing. A single winning creative is exactly that: a single point of failure. When it fatigues (and it will), your entire account performance collapses overnight.

The businesses that maintain consistent results month after month aren't the ones with one brilliant ad. They're the ones with a steady pipeline of different creative concepts, each reaching different buying situations, each given room to find its audience before being replaced by something new.

That's not a creative problem. It's an operations problem. And operations problems are solvable.

Further Reading

• Creative in the Age of Andromeda - Sam Tomlinson's breakdown of how Meta's new ad systems demand genuinely diverse creative
• Meta Andromeda and Creative Diversification: 7 Examples Explained - Jon Loomer's practical guide to Entity IDs and what "different" actually means
• Motion 2026 Creative Benchmarks - Data from 550,000+ ads on what separates winning creative from the rest
• Alchemy: The Surprising Power of Ideas That Don't Make Sense - Rory Sutherland's case for testing counterintuitive approaches
• How Brands Grow - Byron Sharp's foundational research on mental availability and category entry points