You launch a second campaign. Within a week, the first one’s costs have risen and nothing else has changed. This is one of the most common and least diagnosed problems in Meta advertising: you have started competing against yourself in the auction, and you are the one paying the difference.
How the overlap happens
Nobody sets out to build overlapping audiences. It accumulates, usually through some combination of these:
- Broad interest targeting across several ad sets, where the interests are conceptually different but demographically almost identical.
- Lookalike audiences built from source lists that themselves overlap heavily.
- Retargeting windows wide enough to include people who are still inside an active prospecting audience.
- Nested lookalikes — a one percent audience sitting entirely inside the one-to-three percent audience running next to it.
- Past purchasers who were never excluded from acquisition campaigns.
Why the platform does not fully protect you
Meta does de-duplicate to a degree. Within a single ad set it will not run two of your ads against each other for the same impression, and there is some cross-campaign handling.
But once two ad sets have separate budgets and separate optimisation goals, they behave like separate bidders with the same objective. When both want the same person, the auction resolves the way auctions do — the price goes up, and both bidders are you.
Splitting a budget across more ad sets does not spread your reach. Past a point, it just raises your own price.
How to spot it
The signature is fairly distinctive once you know to look:
- CPMs climbing steadily with no seasonal or competitive explanation.
- Frequency rising while reach stays flat — you are showing more ads to the same people.
- An older campaign’s performance declining from the week a new one launched.
- Meta’s audience overlap tool showing high percentages between ad sets you assumed were distinct.
- Two ad sets whose results are eerily similar, because they are effectively talking to the same population.
The structure that prevents it
Most of the fix is exclusions, applied consistently rather than cleverly.
- Always exclude your retargeting audiences from prospecting campaigns. This one is non-negotiable and is skipped constantly.
- Exclude existing customers from everything except retention and upsell campaigns.
- Stack lookalikes so each tier excludes the tighter ones above it — one to three percent excludes the one percent, and so on.
- Consolidate rather than fragment. Fewer, larger ad sets give the optimisation more signal to work with and reduce the surface area for overlap.
Fragmenting for control costs more than it saves
There is a strong instinct, especially with a modest budget, to split into many small ad sets so you can control exactly where money goes. It feels responsible and it is usually counterproductive.
Small ad sets take longer to exit the learning phase, gather less signal, and multiply the chances of two of them chasing the same person. The control is real, but you pay for it in higher costs and slower learning — and on a modest budget, that is exactly the trade you can least afford.
Before you add another campaign, check whether the audience you are about to target is already being reached by one you are running. That single habit will save more budget than most of the optimisations you were planning instead.
