Play · Meta Lookalikes
Primer lets you feed Meta a clean, high-quality seed and a suppression list that says: “these people are off-limits.” That way, the algorithm optimizes within your ICP — not outside it.
Meta’s algorithm is powerful — but lacks B2B signals. It optimizes for conversions but not quality company fit or role. Primer constrains it: you tell Meta exactly who’s in bounds (ICP) and who’s off limits (anti-ICP). You still get Meta’s reach, but within defined B2B guardrails.
Most of our lookalike conversions came from micro and SMB segments. There’s a big difference in LTV when we compare those to enterprise leads. Primer helped us box in Meta’s algorithm — and boost enterprise conversions by 37%
Step 1
Use high-quality converters / customers as the lookalike seed. Sync to Meta.
Tip: See how Verkada rotates lookalike audiences to avoid fatigue — they scaled from 1% to 4% without rapid fatigue.
Product example: in Primer's audience editor, a lookalike seed built from HubSpot customers. Closed-won rates from the CRM show Marketing at 47% by department and Software at 41% by industry, so the seed keeps marketing customers or customers in software: 4,100 companies and 12,400 people, matching 8,600 to 10,500 on Meta, synced as a Custom Audience.
Step 2
Define who’s off-limits (e.g., small companies, students, agencies). Sync to Meta.
Tip: Suppression is what keeps Meta from drifting into cheap clicks / non-buyers — think of it as guardrails for your lookalike.
Product example: in Primer's audience editor, an anti-ICP suppression audience of companies with 1 to 10 employees, students, interns and freelancers, competitors and customers in the United States and Canada: 48.2 million people, matching 31.3 to 41.0 million on Meta, synced as a Custom Audience.
Step 3
Tip: The difference between the two ad sets reveals how much the suppression layer improves quality.
Product example: in Meta Ads Manager, a 1–3% United States lookalike created from the Primer customer seed, then an ad set targeting that lookalike with the Primer anti-ICP suppression audience excluded, narrowing the estimate from about 5.3 million to 3.3 million people, and a duplicate control ad set without exclusions on the same budget, creative and optimization event.
Step 4
In Meta’s Experiments, split spend evenly between:
Keep conversion event constant (e.g., demo booked or signup).
Warning: If your audience or signal volume is low, ad sets may never exit learning mode. In that case, manually optimize based on early cost and conversion trends.
Product example: Meta's Experiments A/B test comparing two ad sets: A, the lookalike with Primer suppression, and B, the lookalike only, each at $200 a day, with a custom conversion as the key metric.
Step 5
Compare performance across:
Ask Primer
Back in Meta’s Experiments, open the test results and compare cost per lead for both ad sets.
In Meta Ads: if there’s no significant difference yet, consider extending the test to gather more data.
Product example: completed Meta experiment results. Ad set A, the lookalike with Primer suppression, brought 185 leads at $65 each; ad set B, the native lookalike, brought 147 leads at $82, on about $12,000 each. Meta estimates a 92% chance of the same result if the test ran again.
Ask Primer
Cheaper leads aren’t the goal. Primer follows both ad sets’ leads into your CRM to see which qualified and moved down the funnel.
Ask Primer: it compares cost per qualified lead and down-funnel conversion rates for both ad sets.
Product example: asked whether the Primer suppression made lookalike leads better, Primer reports a $65 cost per lead against $82, a 44% qualified rate against 28%, a $148 cost per qualified lead against $293, and 17% of qualified leads opening an opportunity against 9%.
Primer tells Meta who’s in and who’s off limits, helping the algorithm focus on your real ICP.
Primer provides a large, structured audience so Meta’s machine learning can optimize within your ICP instead of wandering into irrelevant segments.
Meta’s algorithm is powerful — but lacks B2B signals. It optimizes for conversions but not quality company fit or role, so lookalikes drift into cheap clicks and non-buyers.
Use high-quality converters / customers as the lookalike seed. Primer shows win rates from your CRM, so you can keep the attributes that close best, and syncs the seed to Meta.
A Primer anti-ICP audience of people who are off-limits (e.g., small companies, students, agencies), applied to the lookalike ad set as an exclusion. Suppression is what keeps Meta from drifting into cheap clicks / non-buyers — think of it as guardrails for your lookalike.
Duplicate the lookalike ad set and remove exclusions to create a control, keep budgets, creative and optimization events identical, and use Meta’s Experiments to split spend evenly between Lookalike + Primer Suppression and Lookalike Only (Native Meta). Run ≥14 days or until 50 conversions per ad set.
Yes. Don’t use lookalikes unless your Conversion API (CAPI) is active and sending reliable conversion data back to Meta. Without feedback loops, the algorithm can’t self-correct.
In one comparison, CPL fell from $82 to $65 (–21%), the qualified rate rose from 28% to 44% (+57%) and CPQL fell from $293 to $148 (–49%). Superside used Primer to boost enterprise conversions by 37%.
Build the audience, sync it to every channel, and ask Primer what it drove.
