Short answer: Meta has not added a setting that puts audience selection on the ad itself. Targeting inputs still sit on the ad set. What changed is that Meta's delivery system now reads each creative — the image, the copy, the hook — and routes it to the people most likely to act on it. In practice, the creative has become the targeting.

If you have run Meta ads for more than a year, you have felt this shift even if nobody named it. The interest stack you spent an afternoon building stopped mattering. Two ad sets with completely different targeting started returning near-identical results. A single creative suddenly ate 80% of the budget.

That is not a bug in your account. It is the system working as designed.

What actually changed

Meta's delivery has moved from matching ads to audiences toward matching audiences to ads. Under Advantage+ Audience, the targeting you set is treated as a signal to start from, not a wall. Meta will deliver outside your selected interests when its model predicts a better outcome.

The practical consequence: the strongest lever you have left is the ad itself. Meta cannot deliver to the right person if every creative says the same thing in the same way.

ThenNow
Targeting defined on the ad setTargeting influenced by the creative
Many ad sets, few creativesFew broad ad sets, many creatives
Interests as hard filtersInterests as suggestions
Win by finding an audienceWin by producing distinct concepts
Testing = audience splitsTesting = angle and format splits

Why near-identical creatives get no delivery

Meta groups creatives it judges to be substantially similar and concentrates spend on one of them. This is the single most common reason a batch of "new" ads launches and then flatlines at a few hundred rupees of spend.

The test we apply before launch: if you described two ads to a colleague out loud and the descriptions sounded the same, Meta will treat them as the same. A new colour, a new font, a reordered caption — none of those are a new concept. A different reason to buy is.

How to structure for it

1. Collapse the ad sets

Fewer, broader ad sets give the model more room and stop you splitting conversion signal across a dozen tiny audiences that never exit the learning phase.

2. Test angles, not aesthetics

An angle is a distinct argument for why someone should buy — price, speed, status, risk removal, social proof, a specific objection answered. Build one creative per angle before you build five variations of one angle.

3. Produce on a schedule, not on a crisis

Creative fatigue is predictable. Accounts that ship new concepts every week do not experience the cliff that accounts refreshing quarterly do.

4. Judge creative on its own metrics

Hook rate tells you whether the first second worked. Hold rate tells you whether the middle worked. Cost per result tells you whether the offer worked. A low CTR with a healthy hook rate is a message problem, not a thumbnail problem.

What this does not mean

It does not mean targeting is irrelevant. Exclusions still matter, geography still matters, and retargeting pools still convert at different rates to cold traffic. It means the marginal hour of your week is better spent on a new concept than on a new interest stack.

It also does not mean volume alone wins. Shipping twelve versions of a weak idea produces twelve weak ads. The constraint has moved from media buying skill to creative judgement — which is precisely why creative strategy has become the highest-leverage role on a performance team.

Frequently asked questions

Can you set an audience at the ad creative level in Meta Ads?

Not as an explicit setting. Targeting inputs still live on the ad set. What changed is that Meta's delivery system reads each creative and routes it to the people most likely to convert on it — so the creative effectively decides who sees it.

Is detailed targeting dead?

No. It still exists, but under Advantage+ Audience it behaves as a suggestion rather than a hard filter. Meta can and will deliver outside your selected interests.

How many creatives should a campaign run?

Common industry guidance is eight to twelve distinct concepts per campaign — distinct in angle and format, not in minor visual edits.

Why are my new creatives getting no spend?

Almost always because they are too similar to what is already running. Meta groups near-duplicates and concentrates delivery on one.

A note on sourcing: the structural shift described here is well documented across the industry, but Meta has not published a feature that assigns audiences at the creative level. Specific figures circulating in marketing blogs — similarity-score thresholds, exact winner rates — are vendor claims rather than Meta documentation, so we have not repeated them as fact. For authoritative detail, check the Meta Business Help Center directly.

Running Meta ads and seeing creatives starve for delivery? Book a consultation and we will look at the account with you.