If you've been running Facebook or Instagram ads for your business, you need to stop and read this.

Meta has quietly rolled out the most sweeping overhaul of its advertising system since 2021. And it isn't a tweak. The core tools that small business advertisers have relied on for years -- lookalike audiences, detailed interest targeting, familiar attribution windows -- have either been removed, gutted, or replaced with something fundamentally different.

Here's what changed, what it means, and what you actually need to do about it.

The big one: lookalike audiences are gone

For years, the standard play for Facebook ads was: build a custom audience from your best customers, then tell Meta to find more people "just like them." That's a lookalike audience. It was the workhorse of small business advertising on the platform.

As of February 2026, Meta removed lookalike audiences from Ads Manager entirely.

What replaced them? Meta's Advantage+ system, which uses its own AI to decide who to show your ads to. You provide data signals -- your customer list, website visitors, email subscribers -- and Meta's algorithm takes it from there. You don't define the audience. The machine does.

For some advertisers, this has improved results. Meta's Andromeda AI engine, rolled out globally by March 2026, has gotten notably better at finding buyers across the platform. But the control you had over who sees your ads is largely gone.

Interest targeting got gutted too

Alongside the lookalike removal, Meta eliminated dozens of specific interest targeting categories in January 2026. The granular stacking that used to let you target "people interested in boutique fitness + small business ownership + organic food within 10 miles" is mostly history.

What remains is treated as a suggestion to the algorithm, not a rule. Meta's system can and will expand beyond any interest inputs if it thinks doing so will drive better performance.

The shift is philosophical: Meta is moving from "you tell us who to show ads to" to "give us data and let us figure it out."

New rules on AI content in your ads

Starting this month, if any element of your ad creative was generated or modified by AI -- images, video, copy -- you are required to disclose it. Non-compliance can result in ad rejection, policy strikes, or a temporary hold on your account.

This applies broadly. An AI-enhanced product photo counts. An AI-written headline counts. If you've been using tools like Canva Magic Studio, Adobe Firefly, or any AI image generator to produce ad creative, audit your active campaigns now.

Attribution windows are now much shorter

The window Meta uses to credit your ads for conversions has been compressed. The platform now defaults to a 1-day click-through window for most campaign types. Previously, you could attribute a purchase to an ad if it happened within 7 days of a click.

For businesses with longer buying cycles -- home services, B2B, anything where a customer thinks it over for a week before buying -- this will make your reported conversion numbers look worse. Your ads may still be working. They just won't get credit for it under the new rules.

What to do now

1. Audit your active campaigns. If you have campaigns running lookalike audiences, they've been auto-migrated to Advantage+ predictive audiences. Check that the settings match your intent and that your budget is allocated correctly.

2. Prioritize first-party data. Your customer list, email subscribers, and website visitors are now your most valuable ad assets. Upload them. Keep them updated. The more quality data you feed Meta's system, the better it performs. Consider implementing Meta's Conversions API (CAPI) alongside your pixel for more accurate tracking.

3. Invest in creative quality. With the algorithm handling audience selection, your ad creative is doing more work than ever. One strong video or image will outperform a precisely targeted mediocre ad. Shoot more content. Test more variations. Meta recommends having 15 to 50 active creative variants in Advantage+ Shopping campaigns.

4. Adjust how you measure success. With the shorter attribution window, your cost-per-acquisition numbers will likely increase on paper. Before panicking, benchmark your blended return on ad spend across all channels rather than relying on Meta's in-platform reporting alone.

5. Flag your AI-generated creative. Go through active ads and mark any AI-assisted content. This isn't optional.

The honest take

These changes benefit Meta. An algorithm-first system means advertisers have to trust Meta's judgment, Meta's data, and Meta's reporting. The transparency is lower. The control is lower.

But the counter-argument is real too: Meta's AI has gotten genuinely good at finding buyers. Advertisers who lean into it -- providing strong data signals, investing in creative -- are seeing results. Advertisers who fight it by trying to reconstruct old-style targeting through workarounds are mostly wasting money.

The playbook for Facebook ads in 2026 is: less time defining audiences, more time building great creative and feeding the algorithm clean data.

If you haven't adjusted yet, now is the time. The old approach isn't coming back.

Sources: Audit Socials, Denote, Believe Advertising, Growth HQ