Meta Broad Targeting vs. Interest Targeting: Why the Algorithm Won

Two years ago, handing Meta a completely open audience with no targeting parameters would have seemed like a media buying failure.

Today, that setup is outperforming stacked interest layers on most of the accounts Impremis manages.

The broad targeting versus interest targeting debate on Meta is effectively settled. The data has been consistent and directional for two years. But most eCommerce brands have not updated their account structure to reflect what the data shows — and they are paying for it in higher CPMs, fragmented learning, and campaigns that hit a ceiling they cannot diagnose.

Here is what shifted, why it matters, and what the account structure looks like when you stop fighting the algorithm.

Image brief: Six-row broad vs. interest targeting decision table — Scenario, Recommended Approach, Reason. Interest rows lighter; Broad rows darker. alt: "When to use broad vs. interest targeting on Meta in 2026." caption: "Broad targeting is not the right call in every scenario. Below 100 conversions per week, the model does not have enough signal to outperform manual segmentation."

What Changed in Meta's Algorithm

Meta's machine learning has compounded significantly over the last three years. The combination of more aggregate conversion data, better probabilistic modeling, and the maturation of its AI-driven delivery systems has made the algorithm substantially better at finding likely buyers without being told who to look for.

When you layer in detailed interest targeting, you are not helping Meta find better customers. You are constraining its search to a population defined by proxy signals — people who clicked on outdoor content, or liked cooking pages, or follow fitness accounts — rather than the behavioral and intent signals Meta's models can actually identify.

Interest targeting finds people who have indicated category affinity. Broad targeting finds people who will buy. Those are not the same population, and the algorithm is now good enough to exploit that difference.

The shift is structural. Meta has invested heavily in its AI-driven delivery infrastructure precisely so that manual audience segmentation becomes unnecessary at scale. In 2026, fighting that system with granular interest stacks is one of the most common and most expensive structural errors on new client accounts.

Why Interest Targeting Used to Work

Understanding the history matters because it explains why operators are still running the old playbook.

Before iOS 14, Meta had dense user-level signal data. Interest targeting worked because it was a reasonable proxy for purchase intent in a high-signal environment. A media buyer could stack relevant interest categories and find a reasonably qualified cold audience because the pixel was tracking purchase events cleanly across most of the addressable population.

iOS 14 degraded pixel signal quality significantly. Meta lost visibility into a large portion of purchase events and had to compensate by modeling conversions rather than measuring them directly. That shift changed the calculus on audience targeting.

Broad audiences became more effective — not less — in a degraded signal environment, because broad targeting gives the model the widest possible search space to find patterns in the signal data it does have. Interest targeting in a signal-constrained environment adds noise: you are restricting the audience pool based on proxy categories that were already less reliable, while limiting the model's ability to find buyers who exist outside those interest buckets.

The media buyers who were early to broad targeting in 2022 and 2023 captured a real performance advantage. The ones who are still defending interest stacks in 2026 are defending a model built for a measurement environment that no longer exists.

The Creative Implication Nobody Adjusts For

The shift from interest to broad targeting changes what the creative has to do. Most brands make the account structure change without making the corresponding creative adjustment — and then conclude that broad targeting does not work.

With interest targeting, the audience did significant qualifying work before the ad was even seen. If you targeted "trail running enthusiasts," your creative could assume baseline category familiarity. The ad could lean into specifics that a self-identified category member would recognize.

With broad targeting, the creative is doing all of the qualifying. The ad itself has to attract the right person and repel the wrong one. That is a fundamentally different brief.

This is why broad targeting fails for brands with weak creative infrastructure. It is not that broad does not work. It is that broad exposes exactly how much the creative was relying on audience segmentation to do the filtering.

What strong creative looks like for broad audiences:

Hooks that filter. A hook framed as "if you deal with X, this is for you" does qualifying work. A hook framed as "you need to see this" does not. Broad audiences need the hook to identify who should keep watching — not just interrupt the scroll.

More specificity in the body, not less. The person seeing your ad on broad has no pre-qualified context. They need enough information within the ad to decide whether this is relevant to their situation.

Problem-specific hooks over product feature hooks. In internal testing on broad campaigns, we run three to five hook variants per creative concept, each addressing a different pain point angle, and let the algorithm identify which pain point resonates with the widest purchaser segment. That process is only possible if the creative is doing the qualifying work rather than relying on audience definition to narrow the field.

UGC formats tend to outperform polished brand creative on broad audiences because they feel native to the feed and carry social proof without explicitly claiming it. A real-sounding account of a specific problem and a specific outcome qualifies viewers more effectively than a produced product spot — because it mirrors how people actually talk about the category to each other.

Account Structure for Broad-First Campaigns

The structural implications of moving to broad targeting are significant and frequently mishandled.

Campaign consolidation

Broad targeting works best when each campaign accumulates maximum signal data. Running five separate campaigns with fragmented budgets forces five separate learning phases on five separate signal pools. Two consolidated campaigns with more budget concentrated in each consistently outperform the fragmented structure.

The consolidation principle applies at the ad set level too. Multiple ad sets with overlapping broad audiences split signal data and force the algorithm into redundant learning cycles. One ad set per objective per campaign, with creative variation managed at the ad level, is the structure that consistently performs best for broad-targeting accounts above the signal threshold.

Advantage+ Shopping Campaigns

For eCommerce accounts in 2026, Advantage+ Shopping Campaigns are the logical extension of the broad targeting thesis. ASC gives Meta full control over audience, placement, and optimization simultaneously. For brands with a strong creative library and clean conversion infrastructure, it regularly outperforms manually structured campaigns.

The tradeoff is control. ASC does not exclude audiences cleanly, which creates retargeting overlap and complicates attribution. Running ASC alongside standard retargeting campaigns requires monitoring for conversion cannibalization and adjusting the attribution methodology to account for it — otherwise the retargeting ROAS numbers will look artificially strong.

When interest targeting still has a role

Broad does not win in every scenario. There are specific conditions where interest targeting adds value:

| Scenario | Recommended Approach | Reason | |---|---|---| | New account, under 50 conversions/week | Interest targeting | Broad needs signal volume to optimize effectively | | Niche product with narrow total addressable market | Interest targeting | Broad may waste significant budget on irrelevant impressions | | Established account, 100+ conversions/week | Broad targeting | Algorithm has sufficient data to outperform manual segmentation | | Scaling proven creative to new audiences | Broad targeting | Removes audience ceiling from a proven creative system | | Testing new creative concepts | Broad targeting | Eliminates audience as a variable in the creative test | | Short-term seasonal or event-driven campaign | Interest targeting | Contextual relevance can improve short-term efficiency |

The threshold I use: 100 purchase events per week at the account level. Below that, the model lacks enough signal for broad targeting to be reliable. Above it, broad consistently outperforms interest stacking.

Attribution Complexity on Broad Campaigns

Broad targeting creates a specific attribution challenge worth understanding before restructuring at scale.

Interest-targeted campaigns provide some audience-level traceability in attribution reports. A conversion from a specific ad set tells you something about who bought. Broad removes that segmentation layer — your attribution becomes more dependent on creative-level signals and less granular at the audience level.

This affects how the Meta vs. GA4 attribution gap behaves. On broad campaigns, Meta's reach is wider, view-through events increase, and more of the conversion path happens off-platform — all of which widen the gap between Meta-reported ROAS and GA4-reported ROAS. See why the Meta and GA4 discrepancy is structural for the mechanics that cause this divergence to grow as reach expands.

The resolution is consistent: use MER — total Shopify revenue divided by total ad spend — as the primary efficiency metric rather than platform-reported ROAS. Build your MER dashboard before shifting to broad at scale. You need a clean read on business-level performance to evaluate whether broad targeting is driving incremental purchases or whether you are seeing more attribution credit flowing to Meta as reach expands without a corresponding increase in actual revenue.

When broad campaigns are working, MER improves. When they are not, MER stays flat or declines even as Meta-reported ROAS looks acceptable — because view-through attribution is claiming conversions that other channels and organic behavior produced.

How This Changes the Media Buyer's Role

The practical implication for team structure is significant.

When interest targeting was dominant, media buying skill was substantially about audience construction — knowing how to layer interests, build lookalikes, and segment cold audiences in ways that drove efficiency. That was the primary craft.

With broad targeting, audience construction is largely automated. The media buyer's value shifts toward creative judgment, bid strategy, and structural decision-making. Can they identify when a creative concept is fatiguing before CPMs spike? Can they read auction health signals and connect them to creative performance? Can they manage budget allocation across a consolidated account structure without fragmenting signal?

These skills are harder to hire for than interest stacking — and they are also harder to evaluate in interviews. Operators who built their performance teams around the old model need to reassess what they are testing for when hiring and how they are evaluating ongoing media buyer performance.

The analyst function also becomes more important, not less, in a broad-targeting environment. Someone needs to reconcile Meta's attribution data with GA4, track MER weekly, and flag when broad is expanding reach without expanding incremental revenue. That diagnostic layer is what separates brands that scale confidently from brands that keep increasing spend and find the efficiency gains smaller than the cost increases.

FAQ

Should we pause interest targeting entirely and move everything to broad at once? No — restructure progressively. Start by shifting one campaign type to broad (typically prospecting) while keeping existing interest-targeted campaigns running as a performance baseline. This gives you a direct comparison over 30 days and prevents a full account disruption if broad underperforms for your specific account conditions. Scale broad progressively as the data validates the shift.

We are below 100 conversions per week. What can we do to reach the threshold faster? Consolidate campaigns and ad sets to concentrate signal data. Running fewer, larger ad sets with more budget per ad set accumulates the learning signal faster than running many small ones. Consider temporarily pausing underperforming ad sets to concentrate budget in the highest-converting ones until the account crosses the threshold.

How does broad targeting interact with creative fatigue differently than interest targeting? Broad audiences are larger, which extends creative lifespan — but only up to the point where the responsive segment within the broader audience has been exhausted. See the creative fatigue decay signals for the four leading indicators that apply equally to broad and interest-targeted campaigns. The thumbstop rate and 3-second view rate signals are actually more reliable on broad targeting because they are not confounded by audience-level segmentation effects.

Is Advantage+ Shopping the right move for every eCommerce account? ASC performs best for accounts with strong creative libraries (multiple formats, multiple angles, regular refresh cadence), clean conversion tracking, and products with broad appeal. It underperforms for niche products with narrow audiences, accounts with weak creative infrastructure, and brands with complex offer structures that require specific creative-to-landing-page consistency. Test ASC as an additional campaign type before replacing manual campaigns entirely.

Closing

Broad targeting is not a shortcut. It is a more demanding system that requires better creative, cleaner account structure, and more rigorous measurement to function correctly.

The brands still running stacked interest layers in 2026 are not being conservative or data-driven. They are defending a model built for a signal environment that no longer exists — and leaving reach, efficiency, and scale on the table while the algorithm waits for room to do its job.

Give the algorithm the audience signal. Build creative that does the qualifying work the audience segmentation used to do. Measure performance at the business level, not the platform level.

That is the account that scales.

Keep reading

Pieces I've written on related topics that pair well with this one:

• How to Use Meta Advantage+ Without Losing Control — Meta Advantage+ can generate strong ROAS numbers that hide margin problems. Here's the four-part structure for running it without losing visibility.
• What Actually Works on Meta in 2026: A $100M+ Playbook — Across $100M+ in personal Meta spend and $250M+ at Impremis, here's the creative format playbook that's working post-Andromeda.
• Long-Form Ads Are Working on Meta. Volume Is Still a Trap — Why 5-minute and 14-minute ads are outperforming on Meta, and why producing 100 ads a month is the wrong response to it.
• Meta Doesn't Care About Your Margin: Take Back Control — Meta optimizes for what it can see. Your business runs on what it can't. Here's the three-lever system I use across $250M+ in spend.
• The Creative Fatigue Playbook: Predict When a Meta Ad Is Dying Before It Kills Your ROAS — Meta ad creative fatigue is predictable — if you know which signals to watch.