Meta's interest and behavior targeting was the foundation of Facebook advertising for over a decade. You defined your audience by demographic and psychographic categories — age, gender, location, interests, behaviors — and Meta delivered your ads to matching users. In 2026, this model has been substantially disrupted by iOS 14.5 signal loss, interest taxonomy degradation, and Meta's systematic shift toward AI-based optimization. Understanding what changed — and what now works — is the central strategic question in Meta advertising.
Why interest stacks underperform. Interest targeting has two structural problems. First, the interest taxonomy is based on user activity on Meta's platforms — pages liked, content engaged with, groups joined — which is an imperfect proxy for purchase intent. A user who liked a cycling page five years ago is still tagged as interested in cycling regardless of current behavior. Second, narrow interest stacks create small audience pools that starve the algorithm of optimization data. An ad set targeting an audience of 80,000 people doesn't generate enough traffic to accumulate meaningful purchase signal, limiting how effectively the algorithm can optimize for conversion.
The signal regime shift after iOS 14.5. Before iOS 14.5, Meta's algorithm could observe most of what users did on advertisers' websites — product views, add-to-carts, purchases. This behavioral data was the foundation for both retargeting and the modeling that powered interest targeting optimization. After iOS 14.5 ATT, approximately 40–60% of this on-site signal became invisible for iOS users who opted out. The algorithm compensated by relying more heavily on on-platform signals and lookalike modeling from existing customers, reducing the relative effectiveness of intent-based interest targeting.
What 'broad targeting' actually means. Broad targeting in Meta Ads means running ad sets with minimal audience constraints — typically age (18+) and location only, no interest or behavior stacking. You are not targeting no one; you are giving Meta's algorithm maximum flexibility to find users across its full data model. This approach works when you have sufficient pixel conversion signal (50+ purchase events per week), your creative is specific enough to self-select the right audience, and your offer has meaningful mass-market appeal.
The creative-as-targeting principle. In broad targeting environments, creative selection substitutes for audience selection. A specific, detailed ad that leads with 'for Google Ads managers spending 3+ hours per week on account review' self-selects the right audience through content relevance without interest targeting to narrow distribution. This is conceptually different from traditional advertising: in broad targeting, the message selects the audience rather than the audience selection governing the message.
Testing broad vs. narrow: the right methodology. Run a Meta A/B test with identical budget and identical creative: one ad set using your current interest stack, one using broad targeting (age and location only). Let it run for at least 2 weeks with a target of 50+ conversions in each arm. Evaluate on cost per purchase, not click-through rate or view-attributed ROAS. In most consumer categories, broad targeting outperforms interest stacks for cold audiences when sufficient pixel data exists. In niche B2B categories, test results vary more.
Interest targeting still has valid uses. Interest targeting remains appropriate in specific scenarios: small geographic markets where broad targeting dilutes the audience excessively; cold audiences in accounts with fewer than 50 purchase events per week, where the algorithm needs guidance; and B2B categories with specific professional attributes — job titles, industries, company sizes — where the interest proxy is more precise than it is for consumer categories. These are deliberate exceptions, not the default.
Detailed Targeting Expansion. Meta's Detailed Targeting Expansion option (default in many campaign types) allows the algorithm to serve outside defined interests when it predicts conversion opportunity. With expansion enabled, a narrow interest ad set becomes a broad-ish ad set with the interest stack as an audience suggestion rather than a hard constraint. Most accounts perform better with expansion enabled — it captures the efficiency benefit of broader optimization while keeping the interest stack as a starting signal for the learning phase.
Meta's trajectory toward AI-managed campaigns. Meta's product direction is consistent: Advantage+ Shopping Campaigns for e-commerce, Advantage+ App Campaigns, Advantage+ Audience for lead generation, and Advantage+ Creative all remove manual configuration decisions and replace them with AI optimization. This is not a temporary feature — it reflects where the genuine data advantage lies. Adapting means investing in creative quality, pixel signal quality, and catalog feed quality as the primary optimization levers instead of audience refinement.
Digital Face monitors your Meta audience performance, identifying ad sets where interest targeting produces CPAs significantly above campaign average, surfacing audience overlap between ad sets, and tracking weekly conversion signal volume to assess whether your account has sufficient pixel data for broad targeting optimization to be effective. Free plan at digital-face.nl, no credit card required.