Audience Targeting

Detailed Targeting vs Lookalike Audiences: Manual Criteria vs a Modeled Match

In short: Both are ways to define a prospecting audience without needing your own conversion history to retarget, but they get there through different logic. Detailed Targeting is you stacking interests, behaviors, and demographics you believe describe your buyer, filtered by include and exclude rules you set directly. Lookalike Audiences skip the guesswork by having Meta model resemblance to an actual seed list of your best customers, sized by a percentage you choose. The practical split is whether you have a seed worth modeling: with no conversion data, Detailed Targeting is your only option; with a strong seed, a Lookalike usually outperforms your best guess at interests. Chooser: use Detailed Targeting before you have customers to model, and switch toward Lookalikes once you have enough of them to seed one well.

By the AdFlint research team · Fact-checked against current Google and Meta platform behavior · Last reviewed July 2026

Detailed Targeting

Meta's manual layer of interests, behaviors, and demographic attributes used to narrow an ad set, now largely displaced by Advantage+ audience as the default.

You stack criteria with include and exclude logic, and Meta matches users against profile and activity signals. It still matters for audiences the delivery system cannot infer, for regulated categories, and for exclusions. Two things trip people up: options are retired without notice, and targeting expansion - now folded into Advantage+ audience - means your selections function as a suggestion rather than a fence.

Full definition

Lookalike Audiences

Meta audiences modelled on a seed list, matching users who resemble your source on the signals Meta holds, sized as a percentage of a chosen country.

You pick a source - purchasers, high-value customers, engaged visitors - and a percentage; smaller percentages mean closer resemblance and less reach, larger ones the reverse. A value-based source lets Meta weight by revenue. Their edge has narrowed as broad targeting and Advantage+ audience improved. The persistent error is seeding from a large low-quality list, which models the wrong person very precisely.

Full definition

Side by side.

The differences that actually change what happens in your account.

	Detailed Targeting	Lookalike Audiences
What defines the audience	Criteria you select directly - interests, behaviors, demographics.	Statistical resemblance to a seed list, computed by Meta.
Data required to start	None - works on a brand-new account.	A seed audience of workable size, drawn from your own customers or a Custom Audience.
How precisely you control it	Very - you pick and combine each criterion explicitly with include and exclude logic.	Indirectly - you control the seed and the percentage size, not which individual traits matter.
Time to build	Immediate, since it is just a category selection.	Takes time to model and needs periodic refresh as the seed changes.
Decay mechanism	Options get retired without notice as Meta updates its category catalog.	Model quality drifts if the seed list goes stale or shrinks.
Effect of Advantage+ Audience being on	Inclusion criteria drop to a suggestion; exclusions still bind.	The Lookalike itself, used as inclusion, also drops to a suggestion under Advantage+ Audience.
Value weighting option	None - all criteria are treated equally regardless of predicted value.	Available if the seed list includes a value column, biasing the match toward high-value resemblance.

What actually separates them.

Detailed Targeting needs no seed data and works immediately; Lookalike Audiences need a workable seed list before Meta can model anything, which makes them unusable on a data-free account.

Detailed Targeting gives you direct, inspectable control over which criteria apply; Lookalike Audiences give you only indirect control through the seed and the percentage size, not the traits Meta actually weighs.

Detailed Targeting decays through Meta retiring interest and behavior categories; Lookalike Audiences decay through the seed list going stale or shrinking, a completely different maintenance problem.

A value-weighted seed lets a Lookalike bias toward your best customers specifically; Detailed Targeting has no equivalent, since it treats every selected criterion the same regardless of value.

Both lose inclusion-side enforcement under Advantage+ Audience, but a Lookalike still carries value from its seed's statistical resemblance even as a mere suggestion, while a Detailed Targeting interest carries much less specific signal once downgraded.

Which one should you use?

Use Detailed Targeting when

You have no conversion history or Custom Audience yet to build a seed from.
You need precise, inspectable control over which interests or behaviors are included and excluded.
You are testing a hypothesis about who your buyer is before you have data to confirm it.
You are in a category where regulated targeting rules require explicit, auditable criteria rather than an opaque statistical model.

Use Lookalike Audiences when

You have a sufficiently sized seed list of purchasers or high-value customers to model from.
Your Detailed Targeting guesses have plateaued and you want a data-driven alternative rather than more interest stacking.
You want a controllable resemblance dial - tighter for precision, looser for volume - rather than a fixed category list.
You have a value column in your seed data and want the match biased toward your best customers, not just any converter.

Common questions.

Can a Lookalike Audience replace Detailed Targeting entirely?

Only once you have a seed worth modeling. Before that, Detailed Targeting is the only one of the two that can run at all, since a Lookalike needs an actual customer list or Custom Audience to build from.

Why does my Lookalike Audience keep changing size?

It is modeled off your seed list and updates as that seed changes, so a growing or shrinking customer base shifts the Lookalike along with it. A sudden change usually traces back to the seed source - a Custom Audience that expired out of its retention window, for example - rather than the Lookalike feature itself.

Is it worth stacking many Detailed Targeting interests to approximate a Lookalike?

No. Interests combine with OR logic and stacking many of them tends to produce a broad, blurry audience rather than anything resembling a real customer profile. If you have enough customers to model, a Lookalike will do a better job of finding similar people than a long interest list ever will.

Do both lose accuracy once Advantage+ Audience is turned on?

Both lose their inclusion-side filtering the same way - the criteria or the Lookalike become a suggestion rather than a fence. Exclusions built from either still hold fully regardless of audience mode.

Or stop choosing between them.

AdFlint picks the setting, writes the ads, and keeps optimizing inside the Google and Meta accounts you already own.

Try AdFlint free See how it works

Related comparisons

All comparisons