What the learning phase actually needs
Ad platforms use an initial period to learn who converts, and they need a minimum volume of conversion events to do it. Meta's stated guidance is around 50 conversions per ad set per week; Google's smart bidding has comparable thresholds. Below that, delivery stays volatile and cost per result is unstable.
The critical detail is that this threshold is per ad set, not per account. An account generating 400 conversions a month sounds healthy until it's split across twenty ad sets, at which point each one is seeing five a week and none of them ever stabilise.
Every structural split — by audience, by placement, by creative, by geography — divides the same conversion volume again. Teams build these structures for reporting clarity and control, then wonder why performance is erratic.
Divide your monthly conversions by your number of ad sets, then by four. If the result is under 50, that's your problem and no bid adjustment will fix it.
Why segmentation feels right and usually isn't
Splitting audiences into separate ad sets feels like control. You can see performance per segment, allocate budget deliberately, and pause what isn't working. It's the structure that made sense when platforms had far less signal and manual optimization genuinely added value.
That era ended. Modern delivery systems find pockets of performance within a broad audience more effectively than a media buyer partitioning it in advance — provided they have enough conversion data to learn from. Pre-segmenting is now usually taking a decision away from a system better equipped to make it.
The uncomfortable part is that consolidation reduces visible control and reporting granularity. You lose the per-segment breakdown. What you gain is ad sets that actually exit learning, and stable cost per result is worth more than a report showing why an unstable one was unstable.
When segmentation is still correct
Consolidation isn't universal advice. There are genuine reasons to separate, and they share a characteristic: the economics differ, not just the reporting.
Materially different conversion values
If one product line contributes three times the margin of another, they need different targets. Mixing them means the algorithm optimizes toward volume rather than value.
Separate budget accountability
When distinct business units fund distinct campaigns, separation is an organisational requirement rather than a performance choice. Accept the cost knowingly.
Genuinely different creative and message
A campaign for a new market with different messaging needs its own learning. Forcing it into an existing structure teaches the algorithm the wrong thing.
Regulatory or compliance boundaries
Regulated products, age-restricted categories and geographic licensing limits sometimes require hard separation regardless of what performance would prefer.
How to consolidate without losing the plot
Consolidation done carelessly loses information you actually need. Done properly, you keep the visibility and gain the volume.
Move segmentation from the ad set level into the creative and the reporting layer. Different audiences can be addressed by different creative inside one ad set, and platform breakdowns still show you how each performed — without splitting the conversion signal.
Then use campaign budget optimization rather than fixed ad set budgets, so spend flows toward what's working rather than being locked into a split you decided in advance.
What to expect when you do it
Consolidated campaigns re-enter learning, so expect two to four weeks of elevated volatility before performance stabilises. Teams that consolidate and judge the result after ten days conclude it failed and revert, which is the most common way this change gets abandoned.
Set the evaluation window before you make the change, and communicate it upward. If leadership is watching a daily dashboard, a fortnight of noise after a structural change looks like a mistake unless someone said in advance that it would happen.
The gain is usually visible by week four or five as cost per result settles at a lower level than the fragmented structure ever managed. If it hasn't by week six, the constraint was something else — creative, offer, or landing page — and consolidation just made that clearer.
Terms used in this piece
We do this for a living
If you'd rather not build this yourself, these are the services where it lives.