Social campaigns used to be built by stacking interests, behaviors and demographic filters into a narrow audience. Most large advertisers now do far less of that, and the reason is structural.
Manual targeting was a substitute for weak signals
Early social platforms could not reliably predict who would convert, so they let advertisers supply that judgment through interest and behavior selections.
The advertiser was effectively doing the modeling by hand, encoding a theory about the customer into a set of checkboxes and hoping the platform's labels matched reality.
Those labels were always approximate. An interest tag records an inferred affinity, not an intention to buy, and the gap between the two was invisible in the interface.
Conversion feedback made the system a better guesser
Once advertisers began sending conversion events back to the platform, the system could learn which users actually completed the desired action rather than which ones matched a label.
That feedback is far richer than any set of manual filters, because it reflects observed outcomes across enormous numbers of impressions rather than an advertiser's assumptions.
The platform's delivery model can then explore users no interest stack would have selected, which is where much of the incremental performance now comes from.
Narrow audiences fight the delivery system
A tightly restricted audience limits how many people the system may consider, and a constrained pool tends to produce higher costs as the campaign exhausts the responsive part of it.
Small audiences also slow learning. The system needs a steady flow of conversion events to refine its predictions, and a narrow pool delivers those events slowly or not at all.
Frequency then climbs, since the same limited group sees the campaign repeatedly, and creative wears out faster than it would against a wider audience.
Broad does not mean unmanaged
Broad targeting shifts the advertiser's work rather than removing it. The controls that still matter are the conversion event chosen, the budget structure and the creative itself.
Creative becomes the primary targeting mechanism, because a message written for a particular customer will attract that customer and be ignored by others.
Exclusions remain useful too, particularly for existing customers, recent purchasers and audiences the advertiser is legally or commercially required to leave out.
Measurement has to change with the method
Judging a broad campaign by the composition of its audience is misleading, since the system deliberately reaches people no planner would have chosen.
The meaningful comparison is between structures rather than segments: does a broad setup deliver more conversions at an acceptable cost than a segmented one over the same period.
Advertisers who run that comparison honestly, with enough time for the delivery system to settle, usually find the answer depends on how much conversion volume they can generate.