Broad vs Detailed Targeting on Facebook

You duplicate a winning ad set, and there it is again: the detailed-targeting box with a dozen interests stacked inside it. "Digital marketing," "small business owners," a few lookalikes, three behaviors you added eighteen months ago and never revisited. The question that stops you every time: keep the stack, or delete it and let Advantage+ run broad? Half the accounts you've seen do better on broad. The other half convinced someone that the interests are load-bearing.

Short answer: Broad targeting usually wins now because Meta treats detailed interests as a loose suggestion the delivery system can expand past, so stacking them rarely constrains reach the way buyers assume. Hand-picked interests still earn their keep for tiny local radii, regulated audiences, and cold accounts under roughly 50 weekly conversions. Past that, the audience knowledge moves into the creative.

The takeaways

• Meta's own detailed-targeting guidance says the system can deliver beyond the interests you pick, so a 12-interest stack rarely does what a buyer assumes it does.
• Keep manual targeting for three specific cases: a radius too small for the algorithm to explore, compliance-restricted verticals, and accounts under roughly 50 conversions a week.
• Broad delivery doesn't retire audience research. It moves it into the creative, where the buyer's exact vocabulary becomes the steering signal.

Does detailed targeting still control who sees your ad?

Mostly no. Meta's own guidance now describes detailed targeting as a suggestion the delivery system can expand past when it finds better-converting people elsewhere, and the whole Advantage+ audience push is built on that expansion. Your twelve interests aren't a fence around a room. They're a hint the algorithm reads on its way to a much larger pool.

That matters because a lot of buyers still budget as if the stack were a hard boundary. They narrow, narrow again, and quietly make the learning phase harder while the pool the algorithm actually delivers to stays wide. If the interests aren't constraining delivery, all that layering buys you is a smaller sample to learn from and a slower start. The control you think you're exercising is mostly theatre.

When is manual interest targeting still the right call?

Three cases, and they're narrower than the folklore suggests. First, a geographic radius too small for the algorithm to explore: a single store pulling from a ten-mile ring doesn't give broad delivery room to work, so you set the audience yourself. Second, regulated or compliance-sensitive verticals, where you must be able to prove who could and couldn't see the ad. Third, a cold account.

The cold-account case is the one people skip. Meta's guidance puts the exit from the learning phase at roughly 50 optimization events per week, and under that threshold the AI has almost nothing to learn from. A narrower, hand-seeded audience can give a new pixel a cleaner signal before you hand the reins to broad. Treat it as a starter setting you graduate out of once the events pile up.

If the algorithm picks the audience, why research the buyer?

Because the creative is the last lever you fully control, and under broad delivery it does the targeting. When you can't fence the audience, the way you point delivery at the right people is by writing an ad only the right people respond to. A hook in a landscaper's exact words gets ignored by everyone else, and the algorithm reads those early skips and scrolls as the signal for who to keep showing it to. The vocabulary is the targeting now.

That's the honest reason audience research survives the shift to broad. It stops living in the interest box and starts living in the copy. This is the work Adscalr's audience intelligence does: it searches five voice-of-customer sources (Reddit, Amazon, App Store, Google Play, custom forums), pulls the exact 2 to 4 phrase markers people use, and maps them to Eugene Schwartz's five awareness stages, so the concept is built from words your buyer already said. Broad delivery doesn't make that redundant. It makes it the whole job.

How do you test broad against detailed without fooling yourself?

Run them as a real comparison. Same budget across both isn't apples to apples, because broad usually pulls a cheaper CPM and manual bids against everyone for a small pool, so equal spend buys very different sample sizes. Give each cell enough events to clear its learning phase before you read anything, and judge on the outcome that pays you.

The classic split you'll see: Advantage+ wins on cost per lead, manual wins on lead quality. Both are true at once, because the two settings are optimizing for different things. So don't rank them on cost per lead alone. Rank them on cost per qualified lead, or per real sale, over a window long enough that a couple of lucky days can't decide it. If you're mapping which searchers and buyers sit at which awareness stage first, you already have the language to make the broad cell's creative earn its cheaper clicks.

Broad or detailed is the smaller decision. Knowing the buyer well enough to write the ad that broad delivery needs is the one that compounds.