Somewhere along the way, "just make more creatives" became the answer to every paid social problem. Reach flattening? Ship more. CPMs climbing? Ship more. Performance sliding?
You get the picture. The advice gets repeated in every deck, every agency QBR, every LinkedIn thread, until nobody stops to check whether it is still true.
What do you do when performance dips? If like most growth terms, you order for more creatives, then I’ve got bad news for you. The myth of more volume means you’ll eventually win is wrong.
Even worse is that believing more creatives equals more revenue actually costs much more than most accounts realize.
The myth costs you money before you know it
Here’s what actually happens when you push a large batch of creatives into a single ad set.
The algorithm picks roughly two or three creatives to spend behind. The rest receive a token share of impressions and almost nothing else. By the end of the flight, budget has consolidated into those two or three assets and the remainder has collected just enough spend to feel like it was tested, without ever having a real chance.
So what that means is if you ran ten, fifteen, or twenty creatives hoping diversity would surface winners, most of that creative investment is wasted from the moment the campaign launched.
That doesn’t mean that the ideas were bad. But the fact is that volume isn’t what Meta’s algorithm is optimizing for. If you have been told your creative problem is a volume problem, someone sold you the wrong fix.
Why differentiation is the real lever (the Andromeda mechanism)
Since Meta's Andromeda update, released nearly two years ago, targeting as practitioners used to understand it has been functionally replaced. Before, we’d used to hand-pick who saw each ad and stack interests, demographics, and lookalike audiences to build a tightly defined target group, then let the ad run to just that group. Now, audience specification only operates as a signal, a directional suggestion that the system weighs against everything else it knows, not the hard control it once was. Post update, the creative does the targeting.
What that means day to day, and video is the clearest case: Meta reads the hook, the first three to five seconds, and uses it to decide who sees the ad next. Open two creatives with genuinely different hooks and it reads them as pointing at different people and distributes them accordingly. Open them the same way, even with different footage, and it routes both to the same pool.
And it is not only the opening. Even when two hooks genuinely differ, the body of the creative still counts. If it carries the same vibe (the same creator, similar clothing and background, a similar way of speaking and script structure), Meta reads those as similarity signals and clubs the ads together, routing them to the same people. Even one UGC creator running two different scripts can register as a single signal when everything around the words looks the same.
That is how net-new reach actually accumulates on Meta and why reach stalls when it does. The engagement signal in that window, the share of viewers who actually watch those first few seconds instead of scrolling past, plus early clicks, comments, and shares, is what calibrates the next distribution call. Feed it sameness and you get frequency, not reach. Costs rise, performance flattens, and the account owner blames creative fatigue when the real problem is creative homogeneity.
None of this is unique to video. The same grouping runs across every format: two static images built on the same core message and the same visual treatment get read as one entity, and so do carousels that lead with the same angle. Meta weighs the copy and the visual elements together, and when they feel alike, it clubs them and spends behind one, exactly as it does with near-identical video. When you audit for differentiation, look past your video ads to your static images and carousels too. For a deeper look at what the Andromeda shift means for your creative audit, the implications go well beyond creative count.
What differentiation actually means (it is not what most teams do)
Before anyone congratulates themselves for running "differentiated" creative, let’s understand what variation actually means. Because swapping a background color is not differentiation. Neither is rewriting three words of the hook is not differentiation or running the same script through three different creators. If the hook and the angle are identical, the algorithm reads three clones pointed at the same audience.
Genuine differentiation means different hooks, different angles, different structures: one ad opens on a problem, another on a result, another on a format the category hasn't seen. The question is never "how many creatives are in the account." It is "how many distinct audience signals is this account actually sending." Usually the honest answer is one or two, dressed up as twenty.
A handful of genuinely different creatives will out-reach a crowded ad set of near-identical ones. We can say this with confidence because we’ve seen it happen reliably, because the handful gives the system distinct signals to route while the batch collapses into one pool and burns frequency against the same people.
This is probably an uncomfortable message for any team that has been measuring its own health by production volume, but the count was never the point. The spread between them is.
Iteration used to work, but not anymore
There is a companion piece of dated advice worth killing here, because good operators still swear by it. When a creative won, the playbook was to re-cut it. Pull a new angle from the same footage, swap the voiceover, and test a trimmed version. The logic was sound at the time: you had a proven asset, so doubling down on what worked and mining it was lower risk than building something new.
That logic no longer holds, for the exact mechanism described above.
When you feed Meta near-identical iterations of a winning creative, the algorithm has nothing new to route. It maps them to the same audience signal as the original and distributes them into an already-served pool. Net-new reach drops, the asset pool cannibalizes itself, and what looked like a scaling opportunity turned into stagnation and then fatigue.
This shows up in account after account. When we worked with a recruitment client, a single UGC video captured the overwhelming majority of budget and performed at a cost the team considered exceptional.
The obvious next move was to iterate: recut the footage, tweak the hook, spin up volume off the winner. It’s the move almost everyone would make. Yet The iterations never matched the original, and reach growth flattened anyway.
We’re not saying that iteration is always wrong. If you know precisely which element made the original win (i.e. the hook structure, the emotional frame, or the pacing), and you carry that element into genuinely new concepts instead of derivative cuts, you are not iterating in the broken sense. You’re just learning and applying the learning.
But most iteration isn’t that deliberate. Most iteration is cloning an asset and hoping proximity to a winner transfers. It does not, and post-Andromeda the gap between that hope and the reality is wider than it has ever been.
The lever is available in creatives
The playbook is not complicated, which is part of why the volume advice is so infuriating: a handful of genuinely different creatives, distinct hooks and angles, refreshed by applying what each rotation teaches you rather than by re-cutting the same footage. The goal was never volume, but instead a steady supply of net-new audience signals fed into a system that now rewards exactly that.
If your account is running twenty near-identical creatives and performance has plateaued, you don't need more creatives. You need different ones.
Brighter Click offers a free consultation call where we’ll show you where the spend is going and what differentiation would actually look like in your account. Book your free call today.
