Stacked lookalike ad sets

Quick Answer

Stacking lookalikes means combining several lookalike audiences (e.g., 1% purchasers + 1% video viewers + 1% engagers) into a single ad set rather than running them separately. Stacking gives Meta a bigger pool to optimise against and consolidates learning data. It works well for accounts that struggle to hit the 50-conversions-per-week learning threshold per ad set, but it sacrifices granular CPA visibility per source.

Why stack lookalikes

Meta needs around 50 conversions per ad set per week to exit the learning phase and optimise effectively. If you split your budget across five separate lookalike ad sets, none of them might hit that threshold — especially if your daily budget is modest.

Stacking solves this by combining all your lookalike audiences into one ad set. The conversions concentrate, the algorithm has more data, and learning completes faster.

The trade-off is visibility. When everything is in one ad set, you cannot easily see which lookalike source is driving the conversions. You give up granular reporting in exchange for faster optimisation.

Step-by-step setup

1. Build your individual lookalikes as usual. Common stack: 1% purchasers, 1% high-LTV, 1% video viewers, 1% page engagers.
2. Create one new ad set. Audience > add multiple custom audiences. Meta lets you select multiple lookalikes that are OR'd together (a person in any of them qualifies).
3. Add exclusions. Exclude existing customers, recent purchasers, and any audience you do not want overlap with.
4. Set a higher daily budget than you would for a single lookalike — this is where stacking pays off. Aim for at least £100/day so the ad set hits the conversion threshold.
5. Run identical creative across the stack. Stacked ad sets are not the place to test creative — you want all variables clean except the audience pool.
6. Monitor performance for 14 days. Compare against your previous structure (separate ad sets per lookalike).

Real examples

DTC supplements brand, £200/day budget. Was running five separate lookalike ad sets at £40/day each. None hit learning phase exit. Stacked them into one ad set at £200/day. CPA dropped 22% within two weeks because Meta finally had enough conversions to optimise properly.

B2B SaaS, $300/day budget. Stacked 1% purchasers + 1% trial signups + 1% pricing page visitors + 1% demo requesters. CPA was the same as the best individual lookalike (no decline) but volume doubled because the consolidated audience was much larger.

Local services brand. Stacked five lookalikes from different seed sources (purchasers, leads, callers, page visitors, video viewers). Net CPA improved 18%. The brand could not figure out which source was contributing most, but they did not need to — the stack was working.

When to stack

• Daily budget below £200 and you cannot reach learning phase per ad set
• Multiple seed sources of similar quality you want to combine
• Mature accounts where you trust Meta to allocate within the stack
• Account simplification — fewer ad sets to manage
• You care more about overall CPA than per-source attribution

When NOT to stack

• You are testing audiences — you need separate ad sets to compare
• One lookalike is much higher quality than the others — stacking dilutes the strong one
• Budget is high enough to hit learning phase per individual ad set
• You need granular reporting for finance or attribution

Common mistakes

• Mixing wildly different quality seeds (e.g., purchasers AND all visitors). The weak ones dilute the strong ones.
• Forgetting exclusions. Without them, the stack overlaps with other ad sets and frequency spikes.
• Stacking too many audiences (10+). Diminishing returns and harder to debug.
• Stacking different countries. Lookalikes are country-specific. Mixing creates a multi-country ad set with mixed signal.

FAQs

How many lookalikes can I stack?
Meta technically allows many, but 3 to 6 is the practical sweet spot. Beyond that, you lose the ability to manage the audience.

Will stacking cause audience overlap?
Within the stack, yes (a person in two of the lookalikes will only see the ad once — that is fine). Across separate ad sets, you need exclusions.

Should I stack 1% with 5%?
Usually no. Stack same-percentage lookalikes from different seeds. Mixing percentages creates uneven signal.

Can I stack lookalikes with interest audiences?
Technically yes, but it muddies the signal. If you want both, run them in separate ad sets.

Does stacking work with Advantage+ Audience?
Yes. Provide the stack as the suggestion set, and Advantage+ will use them as signals.

Test creative as carefully as you stack audiences

Your stacked ad set is hitting a diverse audience — your creative needs to land for all of them. Pix-Vu tests creative against multiple audience segments at once so you ship ads that perform across the entire stack. Try Pix-Vu free at https://pix-vu.com.