Lookalike vs. Broad Targeting on Meta Ads: What Actually Works in 2026?

TL;DR: Broad targeting with strong creatives and CAPI data outperforms Lookalike Audiences in 70% of our DACH accounts. Lookalike still wins for niche products, B2B, and budgets under €2,000/month. The real insight: creative quality and tracking accuracy drive 80% of your results. Targeting is the last 20%.

1% Lookalike. Purchase-based. 180-day retention window. That was the targeting setup a Vienna fashion brand inherited from its previous agency in January. Looked solid on paper. In practice, CPA sat at €34.

We launched a parallel Broad targeting campaign — same creatives, same budget, no audience restrictions beyond country and age 18+. Two weeks later, the Broad campaign’s CPA was 18% lower than the Lookalike campaign. Not because Lookalike targeting is inherently broken, but because the rules of the game have changed.

That’s what this article is about. Not theory — real numbers from 30+ account takeovers across the DACH region.

How do Lookalike Audiences actually work in 2026?

Lookalike Audiences are built from a source audience — typically purchasers or high-value leads — and Meta’s algorithm finds statistically similar users. The concept has been on the platform since 2013 (source: Meta Business Help Center, 2024). But the mechanics have fundamentally shifted since iOS 14.5.

Before 2021, Meta had access to a dense web of cross-app and cross-site data. Every purchase, every cart addition, every product page visit fed the algorithm. Lookalike Audiences were a precision instrument because source data was comprehensive.

In 2026? App Tracking Transparency has shredded that data layer. Depending on vertical and geography, 75–85% of iOS users opt out of tracking (source: Flurry Analytics / data.ai, 2024). Your Lookalike source audience is now a fraction of your actual buyers. Meta builds statistical similarities from an incomplete sample — and performance reflects it.

This doesn’t mean Lookalike Audiences are dead. It means they’re no longer the default best option.

Why has Broad targeting gotten so much better?

Because Meta engineered it that way. Sounds glib, but it’s accurate.

Since 2022, Meta has invested heavily in machine learning models designed to work without explicit targeting inputs. Advantage+ Audience — the automatic audience expansion system — is the most visible manifestation, but the core delivery system has been rebuilt from the ground up (source: Meta Engineering Blog, 2023).

Here’s what happens in practice: when you run a Broad campaign, Meta doesn’t just show ads to “everyone in the country.” The algorithm analyzes your pixel data, CAPI events, creative signals, and historical conversion patterns to construct an implicit audience. Broad targeting isn’t really “broad” — it’s algorithmic targeting without manual constraints.

The critical point: the better your tracking data, the better algorithmic targeting performs. In our fashion brand case, combining server-side tracking with Broad targeting dropped CPA from €34 to €24 (−30% in 6 weeks). The algorithm knew exactly who to reach thanks to CAPI data and Advanced Matching — better than any manually defined Lookalike Audience could.

How do Lookalike and Broad compare head-to-head?

Here are the ten criteria we evaluate for every targeting decision. Ratings are based on our internal analysis across 30+ DACH accounts (source: Canem Errant, 2026):

Criterion	Lookalike Audience	Broad Targeting	Winner
CPA (e-commerce, avg.)	€28–35	€22–28	🟢 Broad
Learning phase duration	5–7 days	3–5 days	🟢 Broad
Scalability	Limited (source dependency)	High (full inventory)	🟢 Broad
Niche products (CPA)	€40–55	€55–80	🟢 Lookalike
B2B lead quality	SQL rate 28–34%	SQL rate 12–18%	🟢 Lookalike
Budgets < €2,000/mo.	Stable CPA	High CPA variance	🟢 Lookalike
Budgets > €5,000/mo.	Audience fatigue by week 4	Consistent 8+ weeks	🟢 Broad
Tracking dependency	High (source quality)	Very high (CAPI essential)	⚖️ Tie
Creative dependency	Medium	Very high	⚖️ Context-dependent
ASC compatibility	Limited	Natively integrated	🟢 Broad

The pattern is clear: Broad wins in most standard scenarios. Lookalike fights back where the algorithm lacks sufficient conversion data to find the right audience on its own — niche verticals, B2B, small budgets.

When should you still use Lookalike Audiences?

Lookalike Audiences have three clear use cases left in 2026:

1. Niche products with small total addressable markets. With Erkado — an e-commerce retailer for interior doors — Lookalike based on purchasers kept CPA 22% below Broad. Why? The target audience for interior doors in the DACH region is specific enough that the algorithm needs a head start. Lookalike provided that signal.

2. B2B with long sales cycles. We saw this with a B2B software client: Broad targeting generated cheap leads, but the SQL rate was 12%. A Lookalike audience built from SQLs pushed the rate to 34%, and CPL dropped from €89 to €52 (−42%). More on this approach in our B2B lead generation guide.

3. Budgets under €2,000/month. The algorithm simply lacks volume for Broad targeting at this spend level. With 20 conversions per month, Meta has no statistical foundation. Lookalike gives the system a tighter frame to work within, improving efficiency.

When does Broad targeting win?

In most other cases. Specifically:

E-commerce with 50+ purchases per week — the standard for our DACH clients. With sufficient conversion volume and clean tracking, Broad consistently outperforms Lookalike. The reason: more room for the algorithm, no audience fatigue, no source limitations.

Combined with Advantage+ Shopping Campaigns (ASC). ASC uses Broad targeting by default. You define no audience targeting — Meta optimizes automatically. The results speak for themselves: in our ASC guide, we show how CPA dropped from €34 to €24 for the fashion brand. The shift to ASC makes the Lookalike-vs.-Broad debate effectively obsolete for e-commerce accounts with sufficient volume.

When paired with strong creatives. This is the point most advertisers miss: Broad targeting only works with creatives that stop the scroll. If your ad sits at 0.8% CTR, Broad targeting won’t save you — you need the right creative testing framework first. When we switched the fashion brand from branded creatives to UGC, CTR jumped from 0.8% to 2.3%. That’s the lever that makes Broad targeting fly.

How do you A/B test Lookalike vs. Broad correctly?

The most common mistake: running two ad sets against each other in the same CBO budget and “deciding” after 3 days. That’s not a test — it’s noise.

Our testing protocol:

1. Separate campaigns. Test Lookalike and Broad in distinct campaigns, not as ad sets within a CBO. Why? CBO distributes budget unevenly — it favors whichever ad set delivers a cheap conversion first. That biases the comparison.

2. Identical creatives. The exact same ads in both campaigns. No “variant A here, variant B there.” Otherwise you’re testing creatives, not targeting.

3. Minimum 2-week runtime. The learning phase consumes week one. You won’t see stable CPAs until week two. Cutting early isn’t optimization — it’s panic.

4. At least 50 conversions per variant. Anything less is statistically insignificant. At a €25 CPA, you need €1,250 per campaign — that’s the minimum for a valid test.

5. Compare CPA and ROAS, not CTR. CTR is a proxy metric. What matters is in the 10 KPIs that actually drive e-commerce performance.

Does Advantage+ Audience make this entire debate irrelevant?

Short answer: for most e-commerce accounts, yes.

Advantage+ Audience is Meta’s automatic targeting system that treats your audience inputs as “suggestions,” not hard constraints. You can feed it Lookalike or interest signals, but Meta will expand beyond those boundaries automatically when the algorithm spots better opportunities (source: Meta Business Help Center, 2024).

In practice, this means: if you have Advantage+ Audience enabled — which is the default in ASC campaigns — it doesn’t matter whether you start with Lookalike or Broad. Meta overrides your decision the moment data points in a different direction.

Our hot take: The targeting debate is a distraction. In 2026, creative quality and tracking accuracy determine 80% of your results. Targeting is the last 20%. If you’re still spending hours engineering the perfect Lookalike stack instead of investing in better creatives and clean CAPI tracking, you’re optimizing the wrong lever.

This doesn’t mean targeting is irrelevant. It means the sequence matters: tracking first, creatives second, targeting third. Not the other way around.

Key Takeaway: Broad targeting with CAPI data outperforms Lookalike in 70% of our DACH accounts. Lookalike remains relevant for niche, B2B, and small budgets. But the biggest lever is neither Lookalike nor Broad — it’s creative quality and tracking accuracy (source: Canem Errant, 30+ account analysis, 2026).

What should you do this week?

1. Check your tracking status. Before you even think about targeting: what’s your EMQ? Below 7? Invest your time in server-side tracking, not audience experiments. Without clean data, every targeting strategy is flying blind.

2. Set up a clean A/B test. Separate campaigns, identical creatives, 2-week runtime, budget for at least 50 conversions per variant. No shortcuts.

3. Build your creative pipeline. If Broad targeting wins — and in most cases it will — you need a steady supply of fresh creatives. UGC-first, following the 3×2×1 framework. This isn’t a nice-to-have — it’s the foundation that makes algorithmic targeting work.

Bottom Line: Broad targeting delivers an average e-commerce CPA of €22–28 vs. €28–35 for Lookalike, outperforming in 70% of our DACH accounts. For B2B and niche verticals, Lookalike remains the better choice — SQL rate jumped from 12% to 34%, and CPL dropped 42% (Source: Canem Errant, 2026).

Frequently Asked Questions

Is Broad targeting better than Lookalike Audiences on Meta Ads in 2026?

In 70% of our DACH e-commerce accounts, Broad targeting with clean CAPI data outperforms Lookalike Audiences on CPA (€22–28 vs. €28–35). Since iOS 14.5, 75–85% of iOS users opt out of tracking, which degrades Lookalike source data quality. Broad targeting works better because Meta’s rebuilt delivery system constructs an implicit audience from your pixel data, CAPI events, and creative signals — no manual targeting needed. However, Lookalike still wins for niche products, B2B, and budgets under €2,000/month.

When should I still use Lookalike Audiences instead of Broad?

Lookalike Audiences remain the better choice in three specific scenarios. First, niche products with small addressable markets — with Erkado (interior doors), Lookalike kept CPA 22% below Broad because the algorithm needed a head start on a specific audience. Second, B2B with long sales cycles — Lookalike based on SQLs pushed SQL rate from 12% to 34% and cut CPL by 42%. Third, budgets under €2,000/month where the algorithm lacks conversion volume for Broad to work effectively.

How do I correctly A/B test Lookalike vs. Broad targeting?

Run Lookalike and Broad in separate campaigns — not as ad sets within one CBO, because CBO unevenly distributes budget and biases results. Use identical creatives in both campaigns. Run for a minimum of 2 weeks (week 1 is learning phase). Budget for at least 50 conversions per variant (about €1,250 per campaign at €25 CPA). Compare CPA and ROAS, not CTR — CTR is a proxy metric that doesn’t tell you which targeting actually drove profitable conversions.

Does Advantage+ Audience make Lookalike vs. Broad irrelevant?

For most e-commerce accounts, yes. Advantage+ Audience treats your targeting inputs as suggestions, not constraints — Meta automatically expands beyond Lookalike or interest signals when the algorithm spots better opportunities. In ASC campaigns (where Advantage+ Audience is enabled by default), it doesn’t matter whether you start with Lookalike or Broad. Meta overrides your choice the moment data points elsewhere. Focus your energy on creative quality and tracking accuracy instead — these drive 80% of results.

What matters more for Meta Ads performance — targeting or creatives?

Creatives and tracking matter far more than targeting in 2026. Creative quality and tracking accuracy drive roughly 80% of results; targeting accounts for the remaining 20%. When we switched a fashion brand from branded creatives to UGC, CTR jumped from 0.8% to 2.3% — a bigger impact than any targeting change. The correct optimization sequence is: fix tracking first (EMQ > 7), build a creative testing pipeline second, then optimize targeting third.

---

We’ll review your targeting setup in a free 30-minute audit. No slide decks, no sales pitches — just straight talk and actionable recommendations. Request audit →