Broad Targeting in Meta Ads: Why the Algorithm Knows Better Than Your Interest Stack

TL;DR: Broad targeting — no interest selections, wide age range, both genders — has outperformed detailed targeting in the majority of Meta campaign tests since 2023. The reason is Andromeda, Meta's AI ranking system that turned the auction into a creative-quality contest, not an audience-selection contest. If your creative is good, the algorithm finds your buyer. If your creative is bad, no amount of interest stacking saves you. This post explains why broad won, when it genuinely loses, and how Adlibrary slots into the workflow as the creative-quality diagnostic layer that makes broad targeting viable.

---

Why Everyone Was Wrong About Targeting

The dominant mental model of Meta advertising from 2015 to 2021 was surgical precision. Pick the right interests, layer in the right behaviors, exclude the wrong demographics, and your ads would find the right people. Agencies built entire methodologies — and charged retainers — around this idea. Interest stacks became a form of intellectual property.

That model is structurally dead.

It didn't die because Meta removed detailed targeting. It died because Meta rebuilt the auction from the ground up. The Advantage+ audience system, the Andromeda ranking engine, and the collapse of reliable third-party signals post-iOS 14 all converged on the same outcome: the algorithm knows more about who will convert than any interest stack you can manually configure.

This is uncomfortable for agencies whose value proposition was "we know how to target." It's clarifying for anyone who understands where real leverage lives: in ad creative.

---

What Andromeda Actually Changed

In August 2023, Meta announced Andromeda — a deep learning recommendation engine that replaced the previous ranking system across Reels, Feed, and Stories. The announcement from Meta's infrastructure team (about.fb.com) described it as a move from retrieval-based ranking to full neural retrieval. In practical terms: the system scores every ad against every potential impression using richer contextual signals than any manually specified audience constraint.

What does this mean for your campaigns?

Before Andromeda, your interest targeting narrowed the retrieval pool. The algorithm found buyers within your selections. After Andromeda, your interest selections become a floor, not a ceiling — Meta can serve outside them whenever it finds higher predicted conversion probability. This is Detailed Targeting Expansion (DTE), which has been on by default since 2022 and became the dominant mode through 2024.

The practical result: if you run a campaign with "yoga enthusiasts + women 25–40" and broad targeting simultaneously, the broad campaign will typically find yoga-interested women anyway — because the creative signal (yoga mat imagery, wellness copy) teaches the algorithm what signal correlates with conversion. You get the same audience, plus everyone else who matches the conversion pattern regardless of stated interest.

The Meta Advantage+ Audience documentation frames this explicitly: "Advantage+ audience uses signals from your pixel, app, and catalog to find more people who are likely to be interested in your business, beyond the audience you've specified."

---

Performance Data: Broad vs. Detailed Targeting (2024–2026)

The following benchmarks are aggregated from agency reports, platform disclosures, and third-party audits published 2024–2026. Individual results vary by vertical, creative quality, and spend level.

Sources: Common Thread Collective 2024 State of DTC Advertising report; Motion App Benchmark Report Q1 2025; internal agency benchmarks published by Foreplay (foreplay.co/blog); Meta Ads Manager aggregate campaign data.

The CPM differential alone is decisive. Broad targeting accesses more inventory at lower cost because you're not competing in artificially narrowed pools. CPM is the cost of attention — paying 26% less for it is a structural advantage that compounds through CTR, CPA, and ROAS.

---

When Detailed Targeting Still Wins

Broad targeting is not universally dominant. There are specific conditions where interest and behavior selections still produce better outcomes. Be honest about which bucket your situation falls into before defaulting to either approach.

The honest summary: detailed targeting wins in B2B, compliance contexts, geographic specificity, and sub-scale campaigns. For DTC, e-commerce, and any brand spending $500+/day with a mature pixel, broad targeting wins on economics and scale almost every time.

---

The Mechanics of Broad Targeting on Meta

Setting up broad targeting correctly is not about what you add — it's about what you don't add.

Campaign level: Use Advantage+ Shopping Campaigns (ASC) if you're running e-commerce. These are natively broad and leverage the full Andromeda retrieval system.

Ad set level for manual campaigns:

• Age range: 18–65+ (do not narrow unless compliance requires it)
• Gender: All genders
• Detailed Targeting: Leave blank — zero interest selections
• Locations: Your actual target geography (this is the one restriction that doesn't hurt you)
• Placements: Advantage+ Placements (automatic) — let the algorithm find where your buyer is cheapest

What this does: It hands the algorithm the maximum information space. The creative signal — images, video, copy, hook rate signals from your pixel — becomes the targeting mechanism. Every person who watches your video, clicks your link, or converts trains the algorithm on what the buyer looks like. Over time, the pixel builds a model that is more accurate than any manually specified interest stack, because it's built from your actual buyers rather than categorical proxies.

Budget minimums for broad to work: Broad targeting requires conversion volume to exit the learning phase. Aim for 50 conversions per ad set per week as the threshold. Below that, you're asking the algorithm to learn with insufficient data. If you're under this threshold, cold audience detailed targeting can work as a temporary scaffold — but plan your migration to broad as volume grows.

---

How Broad Targeting Changed the Funnel

When targeting was the lever, the funnel optimization playbook was: better audience → lower CPM → better CPA. Audience selection was the first-order variable.

When creative is the lever, the playbook inverts: better creative → better CTR → better conversion signal → algorithm finds cheaper inventory → lower CPA. Creative quality is the first-order variable.

This changes what you optimize and where you spend diagnostic time.

Old model time allocation:

• 60% building and testing audience stacks
• 30% creative production
• 10% reporting

New model time allocation:

• 60% creative production, testing, and learning
• 20% structural setup (campaign architecture, CBO, budget)
• 20% measurement and attribution

The practical consequence: the best media buying teams in 2026 are not spending hours in Audience Insights building interest stacks. They are spending those hours in Adlibrary, studying which ad creatives in their category produce the signals that train the algorithm.

---

Step 0: When Broad Fails, Adlibrary Diagnoses Creative-Not-Targeting

Here is the single most important diagnostic principle for broad targeting: when broad targeting underperforms, the instinctive reaction is to add targeting restrictions. Nine times out of ten, that instinct is wrong. The root cause is creative quality, not audience scope.

This is where Adlibrary becomes the surgical tool.

The Post-Andromeda Logic:

Andromeda ranks ads against impressions based on predicted value — a composite of engagement probability, conversion probability, and advertiser bid. The creative is the primary input to the engagement and conversion probability scores. A weak creative with a high bid loses to a strong creative with a lower bid.

What does "strong" mean algorithmically? It means the creative produces:

1. High hook rate — people stop scrolling in the first 3 seconds
2. High hold rate — people watch more than 50% of the video
3. High click-through — CTR well above category benchmark
4. High conversion signal — the pixel sees a purchase-correlated behavior pattern

When broad targeting fails to exit the learning phase, it's almost always because the creative isn't generating enough of these signals to train the algorithm. Adding interest restrictions doesn't fix the signal — it just shrinks the pool, which makes the learning problem worse.

How Adlibrary Surfaces This:

Adlibrary's saved ads library and AI ad enrichment layer give you the competitive diagnostic you need before spending a dollar.

When you're preparing to launch a broad campaign, the workflow is:

1. Search Adlibrary for ads in your category that have been running for 30, 60, 90+ days. Long-running ads have passed the algorithm's selection pressure — they are converting well enough to stay in rotation. That longevity is a proxy for creative quality.

2. Study the visual language, hook structure, call-to-action patterns, and ad copy frameworks in those long-running ads. This is the signal your algorithm needs to see in your creative.

3. Use Adlibrary's AI enrichment to categorize these ads by creative angle, emotion, format, and audience persona. This gives you a structured map of what's working in your category — not anecdote, not intuition, but pattern-level data from the competitive set.

4. Build your creative to match or out-execute those patterns. Then run broad.

If your broad campaign still underperforms after this, check creative fundamentals — hook rate in your first 3 seconds, visual contrast, offer clarity. Don't reach for interest restrictions.

The moat Adlibrary creates is precisely this: prospecting with broad targeting requires good creative, and good creative requires knowing what "good" looks like in your specific category. Adlibrary is the intelligence layer that closes that gap. Warm audiences and custom audiences built from the high-quality seed traffic that good creative generates will then compound your results.

---

Broad Targeting and the Funnel Ecosystem

Broad targeting is a cold audience prospecting strategy. It doesn't operate in isolation — it feeds downstream systems.

The full architecture:

1. Broad prospecting (cold): Advantage+ or manual broad ad sets surface the creative to the widest relevant pool. Pixel trains on who converts.

2. Lookalike audiences: Meta builds lookalikes from your purchasers, high-value customers, and engaged visitors. These are algorithmically similar to your buyers — seeded from real conversion data, not from interest proxies.

3. Custom audiences: Retargeting warm audiences who engaged with broad prospecting but didn't convert. Video viewers, link clickers, add-to-cart abandoners.

4. Advantage+ Shopping Campaigns: The integrated system that runs broad prospecting and retargeting together, letting the algorithm allocate budget dynamically across both.

When broad targeting is working, each layer strengthens the next. The broad creative finds the buyer. The lookalike seed grows from real conversions. The custom audience retargeting closes the loop. The conversion rate compounds across the funnel.

If you're running detailed targeting instead, you're constraining the top of this system. Your lookalike seeds are smaller, your pixel data is noisier, and your retargeting pool is shallower. The economics degrade at every layer.

---

Budget Thresholds and Scaling

Broad targeting's advantage grows with spend. Here's the practical framework:

$0–$100/day: Detailed targeting or small cold audience lookalikes as scaffold. Broad doesn't have enough conversion volume to learn.

$100–$500/day: Transition zone. Test one broad ad set alongside your best detailed targeting ad set. Measure CPA, ROAS, and learning phase exit. Most accounts see broad match or beat detailed in this range within 2–3 weeks.

$500–$2K/day: Broad targeting typically dominates. Run Advantage+ or CBO with broad ad sets. This is where the CPM advantage compounds most visibly.

$2K+/day: Broad targeting at scale. Advantage+ Shopping Campaigns with multiple creative variations running simultaneously. The algorithm allocates budget dynamically to whichever creative is performing best for each impression opportunity.

The scale ceiling difference is decisive: Detailed targeting ad sets hit frequency walls as the audience saturates. A 500K-person interest audience runs out of new impressions within days at $2K/day. Broad targeting has no practical ceiling — you're running against millions of people, with the algorithm continuously finding the subset most likely to convert on each creative.

This is why ad fatigue hits differently in detailed vs. broad. Detailed targeting fatigues the audience. Broad targeting only fatigues the creative — and that's fully within your control.

---

The Creative Rotation Imperative

If broad targeting means creative is the lever, then creative testing becomes the primary ongoing operation. The question isn't "which audience should I target?" — it's "which creative earns the algorithm's distribution?"

The cadence that works:

1. Run 3–5 creative variations per ad set simultaneously
2. Let the algorithm allocate impressions via dynamic creative or through CBO budget allocation
3. After 7–14 days, identify the winner by CPA and ROAS
4. Rotate out losing creatives, introduce 2–3 new variations
5. Repeat indefinitely

What Adlibrary accelerates in this process:

Before you shoot a new video or brief a designer, use Adlibrary's competitive intelligence to understand what's already working in your category. Long-running competitor ads in your category have survived selection pressure — they are earning distribution. Study their structure: What's the hook? What's the call-to-action? What emotional register does the ad copy use? What creative angle is dominant?

This research eliminates guesswork from creative briefing. Instead of briefing based on gut instinct, you brief based on pattern evidence from creatives that the algorithm is already rewarding. The result is a higher first-batch win rate and a shorter time to finding a profitable creative.

---

Common Broad Targeting Mistakes

Mistake 1: Going broad before the pixel has data Broad targeting is powerful when the pixel has 50+ conversion events per week. Before that threshold, you're running blind. Use cold audience seeds — lookalike audiences from your email list, custom audiences from existing customers — as a bridge.

Mistake 2: Confusing placement restrictions with audience restrictions Restricting placements (e.g., removing Audience Network) is different from restricting audiences. Placement restrictions are often worth testing — particularly removing placements with poor conversion rate. Audience restrictions should be avoided unless compliance requires them.

Mistake 3: Interpreting learning phase failure as targeting failure If your broad campaign doesn't exit the learning phase in 7 days, the instinct is to add targeting restrictions. Don't. Either the creative needs to improve (most common cause) or the budget needs to increase to generate sufficient conversion volume. Check your CPM first — if it's anomalously high, you may have a creative quality penalty. If CTR is below 1%, the hook is failing.

Mistake 4: Running broad without a creative refresh cadence Broad targeting without ongoing creative rotation leads to ad fatigue faster than you expect — because you're reaching your actual buyers faster, at scale. If you're running $1K+/day broad, plan for at least one new creative batch every 2–3 weeks.

Mistake 5: Not using AI ad enrichment before briefing The biggest waste in broad targeting campaigns is building creative that doesn't match the category's conversion patterns. Before any creative brief, spend 30 minutes in Adlibrary identifying what the algorithm is already rewarding in your space. That research pays for itself on the first campaign.

---

Broad Targeting in Specific Contexts

DTC e-commerce: The natural home of broad targeting. High conversion volume, mature pixels, direct-response creative that generates clean signals. Run Advantage+ Shopping with 5–8 creative variations. Let CPA decide the winner.

Lead ads: Broad targeting works for lead generation campaigns, with one caveat: lead quality degrades when the audience is too wide. Monitor lead-to-close rate alongside lead CPL. If quality drops, narrow by geography before narrowing by interest.

Reels ads: Reels inventory is increasingly where broad targeting delivers lowest CPM. Vertical video creative optimized for Reels performs best in broad because Reels inventory is massive and underpriced relative to Feed.

Brand campaigns: Brand awareness campaigns should always run broad. Reach and frequency campaigns don't require conversion signals — they require scale. Broad is the only rational choice.

Dynamic creative: Dynamic creative optimization pairs naturally with broad targeting. Give the algorithm headline, image, body text, and CTA variations. It will serve the combination most likely to convert for each individual impression. The combination of broad audience + dynamic creative is where the algorithm's advantage over manual optimization is most visible.

---

Measuring Broad Targeting Performance Correctly

Broad targeting makes attribution more complex — not less accurate, but harder to read if you're using last-click only.

What to measure:

• 7-day click + 1-day view ROAS: The standard Meta attribution window for broad prospecting. Captures view-through conversions that broad targeting generates at scale.
• Blended ROAS: Total revenue divided by total ad spend, across all channels. Broad prospecting influences organic search, email, and direct traffic — last-click attribution misses this.
• MER (Marketing Efficiency Ratio): Same as blended ROAS but sometimes calculated on contribution margin. The cleanest top-level metric for broad campaign health.
• CAC: Monitor new customer acquisition cost, not just purchase ROAS. Broad targeting can shift buyer mix — watch whether you're acquiring new customers or retargeting existing ones.
• Payback period: If LTV is strong, broader targeting with a slightly higher CAC can still be the right choice. Model payback period before optimizing purely on short-window CPA.

---

Frequently Asked Questions

Q: Does broad targeting work for new accounts with no pixel data?

A: With no pixel history, broad targeting is genuinely risky — the algorithm has no signal to work with. The best approach for new accounts is to start with a seeded lookalike audience built from your email list or existing customer data, or run a small cold audience detailed targeting campaign to build initial conversion history. Once you hit 50 conversions per week, migrate to broad. Don't try to shortcut the seed phase.

Q: Won't broad targeting waste spend on people who will never buy?

A: Less than you think, and less than the alternative. Detailed targeting "precision" is largely theatrical post-Andromeda — Detailed Targeting Expansion means Meta serves outside your selections anyway. The CPM cost of detailed targeting (26% higher on average) is a real penalty you pay for false precision. Broad targeting at a lower CPM with better algorithmic retrieval typically produces lower waste in absolute dollar terms.

Q: Should I still test different audiences with broad targeting?

A: Audience testing matters less post-Andromeda. The productive test surface is creative, not audience. Instead of running five ad sets with different interest stacks, run one broad ad set with five different creatives. Your creative test yields actionable learning — the creative that wins tells you what resonates with buyers. An audience test with broad targeting just tells you that the algorithm found buyers wherever it looked.

Q: How does broad targeting interact with Advantage+ Shopping Campaigns?

A: Advantage+ Shopping Campaigns (ASC) are inherently broad — they don't accept manual audience restrictions at the ad set level. Running ASC is the most optimized expression of broad targeting. The campaign uses all available signals — pixel data, catalog data, behavioral signals — to find buyers without any manual audience specification. If you're on e-commerce and not running ASC, test it against your best manual broad campaign. Most accounts see ASC match or outperform within 2–3 weeks.

Q: My agency says detailed targeting protects brand safety. Is that right?

A: This is a common conflation. Brand safety is about ad placement (which websites or apps your ads appear on), not about which audience receives them. Placement exclusions handle brand safety concerns. Detailed targeting restrictions do not prevent ads from appearing next to objectionable content — that's what placement controls and brand safety tools are for. If your agency is conflating these, it's worth surfacing the distinction.

---

Where Broad Targeting Goes From Here

The trajectory is linear. Meta's system is becoming more algorithmic and less manual with every major update. Advantage+ Shopping, CBO dynamic allocation, AI-generated creative, and the full Andromeda rollout across all placements are all moving in the same direction: toward giving the algorithm maximum latitude, and toward creative quality as the primary competitive differentiator.

The implication for your practice: the earlier you shift from audience-centric to creative-centric thinking, the more durable your advantage. Agencies and media buyers who are still positioning audience expertise as their core value proposition are selling a capability that is being systematically commoditized.

The practitioners who win in the broad targeting era are those who:

1. Build creative testing infrastructure that generates learning systematically
2. Use competitive intelligence (Adlibrary) to understand what the algorithm is already rewarding
3. Measure with blended ROAS and MER rather than last-click attribution
4. Structure campaigns (CBO, Advantage+) to let algorithmic allocation work

The broad targeting era is already here. Andromeda made it the default. The question isn't whether to adopt it — it's how quickly you can build the creative infrastructure to make it work.

---

Adlibrary helps you build that infrastructure. Our saved-ad library and AI ad enrichment layer surface the creative patterns the algorithm rewards in your category — so your broad targeting campaigns launch with signal instead of guesswork. See how it works.