B2B lookalike audiences

“Lookalike audience” means two different things. Ad-platform lookalike audiences(Meta, LinkedIn, Google) are built inside the ad platform and can't be independently measured from outside. A B2B lookalike company list — accounts like your best customers, for outbound and ABM — can. This is an independent benchmark of company lookalike / similar-companies APIs — Ocean.io, Exa, Parallel, Seltz, PredictLeads, Extruct, CUFinder, Discolike, and ZoomInfo — ranked on how relevant the companies each one returns actually are, across 48 B2B seed companies.

[01] evidence

Benchmark evidence and per-seed audit trail

Same seed companies, same judge panel, Precision@10 and Precision@100 where each endpoint supports that depth — plus the per-seed matrix so you can inspect how each vendor behaved on each company.

long listPrecision@100: how relevant a vendor stays a hundred companies deep. The metric for lists that feed automated sequences. 3 vendors cannot appear here: their endpoint returns fewer than 100 results.

Long-list lookalike quality: aggregate Precision@100 by company lookalike API vendor
Rank	Vendor	Precision@100	Avg latency	Relevant / $1 at 100
1	Seltz	63.4%	4,835 ms	12,683
2	Extruct	61.1%	2,361 ms	624
3	Ocean.io	59.5%	2,444 ms	37
4	Parallel	56.9%	2,680 ms	11,383
5	Exa	48.6%	3,014 ms	501
6	Discolike	34.6%	1,164 ms	66

short listPrecision@10: how relevant the top of the list is. The metric that matters when a person works the results by hand and only the first few rows get touched. Every measured vendor appears here, and the only question asked of them is whether the companies are right.

Short-list lookalike quality: aggregate Precision@10 by company lookalike API vendor
Rank	Vendor	Precision@10
1	PredictLeads	89.8%
2	Seltz	84.4%
3	Exa	80.8%
4	Extruct	75.0%
5	Ocean.io	74.3%
6	Parallel	69.2%
7	ZoomInfo	67.1%
8	Discolike	41.9%
9	CUFinder	38.9%

benchmarks/lookalike/lookalike-2026-q348 examples · 9 vendors

audit viewcell = Precision@100; the small label shows P@10N/Avendor not yet run (or returned fewer than K results)click any scored cell to view the companies that vendor returned

#	Company type	Ocean.io	Exa	Parallel	Seltz	PredictLeads	Extruct	CUFinder	Discolike	ZoomInfo
01	Cloud observability and monitoring platforms that offer any of application performance monitoring, log management, security monitoring, or synthetic monitoringDevtoolsseed exampleDatadogdatadoghq.com
02	Restaurant point-of-sale platforms that offer any of payments, online ordering, payroll, or inventory managementHospitalityseed exampleToasttoasttab.com
03	Plumbing and drain services companies that offer any of plumbing, drain cleaning, water cleanup, or home servicesHome Services SaaS / Chainsseed exampleRoto-Rooterrotorooter.com
04	Construction management software that offer any of project management, financials, quality, safety, or contractorsB2B SaaSseed exampleProcoreprocore.com
05	On-demand consumer super-apps that offer any of ride-hailing, delivery, payments, financial services, or merchant servicesLogisticsseed exampleGrabgrab.com

Full routing by workflow, latency, cost, and methodology: the lookalike benchmark →

[02] faq

B2B lookalike audiences — common questions

What is a B2B lookalike audience?

The term means two different things. In advertising, a lookalike audience is a set of ad-platform users (Meta, LinkedIn, Google) that the platform's own model builds from a seed list, for ad targeting only. In B2B sales, a 'lookalike audience' usually means a lookalike company list — accounts that resemble your best customers, used for outbound, ABM, and territory building. Different products, different vendors, differently measurable.

What is the best tool to create a lookalike audience from a customer list?

If you mean ad audiences: upload your list to Meta or LinkedIn — the matching happens inside the ad platform and its quality isn't independently measurable from outside. If you mean a B2B lookalike company list from your customers, that is measurable, and measured: Seltz currently leads on long-list relevance (Precision@100 63.4%), PredictLeads on top-of-list precision (Precision@10 89.8%) — on identical seed companies, scored by a three-model LLM judge panel.

How do I build a lookalike audience from closed-won customers?

Pick your strongest, fastest-closing customers as seeds (a clean short list beats a large messy one), feed them to a lookalike tool, and get back ranked similar companies. For ads, upload the resulting list to the ad platform as a matched/seed audience. The step-by-step version for outbound is the closed-won lookalike play.

Can I use these lookalike lists for ad targeting?

Yes — a measured lookalike company list can be uploaded to LinkedIn (company list matched audiences) or converted to contacts for Meta. Note the split honestly: the list quality here is measured; what the ad platform's own matching does with it afterward is not.