Meta Deleted Interest Targeting in January. What Buyers Use Now

Meta interest targeting disappeared from Ads Manager on January 15, 2026, and the saved audiences buyers had tuned for years went with it. Log in that week and the detailed-targeting menu was thinner, whole interest categories were gone, and there was no warning beyond a changelog line. Custom audiences, lookalikes, and broad targeting still work, but the job of expressing intent has moved off the audience panel entirely. Here is what to build in its place.

Meta removed dozens of detailed-interest categories from Ads Manager on January 15, 2026, and cut manual targeting out of Advantage+ catalog ads outright. What remains is retargeting, custom audiences, lookalikes, broad targeting, and Meta's own AI-driven discovery. With audience selection gone as a lever, targeting intent now has to be expressed through creative variation and first-party signal quality, particularly Conversions API match rate, instead of an interest list.

Meta interest targeting lost its detailed layer

On January 15, 2026, Meta removed dozens of detailed-interest targeting categories from Ads Manager — the menu buyers used to stack "engaged shoppers," "frequent travelers," and hundreds of other interest and behavior signals onto a campaign (eMarketer). The cut sat inside a larger restructuring: that February, Meta folded Manual and Advantage+ Shopping into one unified campaign flow, with AI-driven audience, placement, and budget decisions on by default and merely toggleable, not absent.

The removal wasn't cosmetic. Every advertiser has been running on Meta's Andromeda retrieval layer, the GEM ranking model, and the Lattice auction system since October 2025, whether or not the account opted into AI-driven targeting (Search Engine Land).

Meta's targeting taxonomy always split into three layers: demographic targeting (age, gender, location), behavioral targeting (purchase and device patterns), and detailed interest categories sitting on top of both. The January cut hit that interest-targeting layer directly — Meta's own count is "dozens" of categories, without a public itemized list.

What Advantage+ catalog ads lost outright

Advantage+ catalog ads took the sharpest hit of any campaign type when Meta interest targeting was removed. Manual targeting is gone from catalog ads entirely, not narrowed. Retargeting and custom audiences are the only advertiser-supplied signal left, with Meta's AI discovery filling every row a buyer used to fill by hand (eMarketer, cited above). An account that built its catalog structure around ten or twelve interest-segmented ad sets feeding one product set no longer has that structure to build.

Meta's own reported testing on Advantage+ Sales campaigns claims an average +32% ROAS and -17% CPA against manual-only setups. That is Meta's own vendor-published figure, not an independently verified one, and it is the number Meta is using to argue the trade is worth making. Whether it holds for a specific catalog account is a testing question, not a given — treat it as a hypothesis to check against your own results, not a benchmark to inherit.

The interest targeting controls that survived

What survived the Meta interest targeting cut isn't nothing. Custom audiences built from Pixel, Conversions API, and customer-list data still exist, as do lookalike audiences modeled off them. Broad targeting, with no interest constraint layered on at all, still runs, and for most non-catalog campaign types it now carries more of the delivery load than it used to, precisely because there's no competing interest-stacked ad set siphoning off signal. Neither audience type changed mechanically. What changed is that they're now the primary advertiser-supplied lever instead of one option among several.

Creative is the targeting mechanism now

With detailed interest categories gone as a lever, the variable that moves who sees an ad shifted from audience selection to the ad itself. Creative is functioning as the new targeting layer. A different hook pulls a different slice of Meta's retrieval pool, and so does a different visual style or offer, because the model routes on early engagement signal rather than a declared interest.

Practically, a buyer who used to run one creative against six interest-stacked ad sets now runs six creative concepts against one broad ad set and lets delivery sort the audience. Creative testing cycles resolve faster in the Andromeda era partly because the model isn't also fighting audience overlap between ad sets competing for the same pool — a mechanical reason testing has sped up, not a magic one.

Studying what is already winning in a category beats guessing at a new creative brief from scratch. Scanning in-market creative through something like adlibrary's unified ad search, filtered by platform and format, with a live view of how long a variant has held placement, is a cleaner proxy for "what the algorithm currently rewards" than any interest list ever was.

Run through AI ad enrichment, a scroll through hundreds of competitor ads turns into a tagged, searchable set instead of a manual log nobody revisits.

What first-party signal quality determines

If creative decides who the algorithm shows an ad to, Conversions API increasingly decides how well Meta can judge who actually converted. Custom audiences and lookalikes are only as good as the events feeding them, and a Pixel-only setup already leaks purchase events to ad blockers, in-app browsers, and cookie limits before targeting even enters the picture.

That is why Meta explicitly recommends running Pixel and its server-side event system together, deduplicated by event ID, rather than one or the other (Meta Conversions API documentation).

Event match quality, how cleanly a hashed email, phone number, or external ID lines up with a real Meta account, now separates a custom audience the model can actually use from one that is mostly noise. With interest categories gone, that match quality is close to the only advertiser-controlled input left that changes who a lookalike or retargeting set actually reaches. An account that spent January rebuilding creative but left event matching neglected only solved half the problem.

Rebuilding prospecting without an interest stack

Prospecting built on interest stacks needs a different starting point now, not a smaller version of the old structure.

Step 0: find the angle before touching the campaign. Before building a single ad set, check what is already converting in the category, competitor hooks, offer structures, formats still holding placement, instead of starting from a blank creative brief. Pulling that from adlibrary, or scripting it against the API with Claude Code, turns a guess into a shortlist before spend goes out.

1. Collapse the ad set count. Interest stacking used to justify five or ten ad sets per campaign, each carrying a narrow slice of the audience. One or two broad ad sets now carry more delivery than a fragmented structure, because there's no longer a reason to split signal across ad sets chasing the same buyers.
2. Load creative in volume, not variations. Five to eight genuinely distinct concepts per ad set, not five crops of one concept. The model needs real creative diversity to do the audience-sorting work interest categories used to do.
3. Tighten Conversions API match quality before scaling spend. Fix event coverage and deduplication first. A broad ad set fed by clean events out-targets a broad ad set fed by a half-working Pixel every time.
4. Budget for a rougher learning phase. A structural change this size resets exit criteria across every ad set touched. Check spend and conversion-volume thresholds with a learning phase calculator before reading early CPA as a verdict.
5. Run new accounts through a 30-day cold-ramp plan, not a scaled-down version of a warm-audience structure: a cold audience ramp sequences creative volume and budget differently than a mature account does.

In the catalog accounts we've been watching rebuild since January, the ones keeping five or six creative concepts rotating weekly are holding frequency down without an interest layer to spread reach across ad sets. The ones that just deleted their old ad sets and relaunched one broad set with the same three creatives are seeing frequency climb fast instead, the expected result of removing a spreading mechanism without replacing it.

What to stop doing

• Stop building against ASC/AAC campaign creation on the Marketing API. Full deprecation landed May 19, 2026 — v24.0 blocked new creation in October 2025, v25.0 closed it everywhere in Q1 2026, and Advantage+ campaign objects (ADVANTAGE_PLUS_SALES, _APP, _LEADS) are the only supported path now (ppc.land).
• Stop assuming a submitted asset renders as submitted. Advantage+ Creative has drawn agency complaints for cropping logos, adding music, and rearranging text without a review step, prompting some agencies to withdraw from the feature entirely (reported via mambadigital). Check what actually went live, not what was uploaded.
• Stop faking an interest signal with workarounds. Layering placement, device, or demographic filters to approximate a deleted interest category mostly narrows delivery without adding real targeting precision back.
• Stop treating Meta's +32%/-17% figures as your benchmark. They're Meta's own reported testing number, not an independently verified result for any specific account or vertical.

Frequently asked questions about meta interest targeting

What happened to Meta interest targeting? Meta removed dozens of detailed-interest targeting categories from Ads Manager on January 15, 2026, and cut manual targeting out of Advantage+ catalog ads entirely, leaving retargeting, custom audiences, lookalikes, broad targeting, and AI-driven discovery as what remains.

Can you still target interests on Facebook ads in 2026? Only through what survived the cut — broad targeting with no interest layer, plus custom audiences and lookalikes built from first-party data. The detailed-interest category menu itself is gone for the campaign types affected.

What replaced detailed targeting on Meta ads? Creative variation and first-party signal quality, mainly Conversions API event match rate, now do the work that interest and behavior categories used to do, because Meta's retrieval system routes on engagement signal rather than a declared interest.

Do Advantage+ catalog ads support interest targeting? No. Advantage+ catalog ads lost manual targeting outright in the January change. Only retargeting, custom audiences, and Meta's AI discovery remain as inputs.

Is broad targeting better than interest targeting now? Broad targeting isn't inherently better. It's what's left with the fewest constraints, and the pattern holds up across most non-catalog accounts: it tends to outperform a narrow interest stack specifically because it no longer competes with other ad sets for the same signal pool.

Meta interest targeting is gone, and it wasn't replaced with a better version of itself. It was replaced with a bet that creative variation and clean first-party data carry more signal than a declared interest ever did, and Meta's retrieval stack is built to test that bet at scale. The buyers adjusting fastest aren't mourning the interest stack. They moved the hours they used to spend on audience research into creative volume and event-match hygiene instead.

Rebuilding prospecting without an interest stack starts with knowing what is already working in your category. Start a free trial of adlibrary's ad research, or see pricing.