How audience overlap works
Audiences on Meta are not exclusive clubs. One person can sit in your website visitors list, inside a lookalike built from your customers, and within a broad interest audience, all at the same time. Audience overlap is the share of one audience that also appears in another, and the Audiences section of Ads Manager will compare saved audiences and show you how much they share.
Overlap on its own is not a fault; it is arithmetic. It becomes a problem when separate ad sets are actively targeting those same people. Meta filters part of this out — when two of your ad sets are eligible for the same person at the same moment, generally only one enters the auction — but the filter arrives after the money has been divided, so both ad sets are still funded, still learning separately, and still covering the same ground.
Why audience overlap matters
The first cost is learning. Every ad set needs its own run of results before delivery settles, so two ad sets chasing one pool of people take far longer to get there on the same budget than one would.
The second is frequency, and a small market sharpens it. In Nepal, or in a single city anywhere, the number of people who match a narrow interest and are worth advertising to is limited, so three or four apparently different audiences are often the same crowd with different labels — and that crowd sees your ads far more often than the plan assumed.
The third is reporting. When the same person can be reached from several ad sets, comparing those ad sets tells you less than it appears to. You are not testing audiences against each other; you are testing which ad set reached them first.
Where audience overlap goes wrong
Stacking several lookalike audiences of neighbouring sizes is the usual source, because the larger one contains the smaller ones almost entirely. Running retargeting beside prospecting with no exclusion is the next, so people who already visited get bought twice, once at a cold price.
Splitting a small country by city creates overlap of a different kind. Location targeting is not precise enough to keep the edges apart, and people cross those edges during an ordinary working day.
How to act on it
Check overlap before launch rather than after a month of confusing results, and compare the audiences you genuinely intend to run at the same time. Where two share most of their members, keep one.
Then set exclusions deliberately: existing customers out of prospecting, recent buyers out of retargeting. Where an audience simply duplicates another, delete it rather than excluding it, because the excluded ad set still costs you budget and attention. Consolidating into fewer, broader ad sets usually solves the overlap and the learning problem in one move. If an account has accumulated audiences over several years, an audit of the Meta ads account is the practical way to find which are duplicates.
