We sampled 10,000 domains from the 719,358-domain IAB Advertising vertical and ran them through both pipeline passes. The sample alone surfaced 673 eligible agencies, each with a complete 23-signal extraction. Below: the full pipeline funnel and the 20 highest-scoring candidates with every visible signal, confidence score, and quoted evidence.
This is a sample run — 10K drawn at random from the full IAB Advertising vertical. The sample alone produced 673 fully screened agencies. A full-vertical run would process all 719K domains.
Total domains categorized under IAB “Advertising” across the open web.
Uniform random sample for this pipeline demonstration. A full engagement processes the entire vertical.
Homepage fetched, AI classifies: operating company vs. parked/dead, country, team size, independence, subvertical.
PE-owned entities, excluded verticals, and non-independent firms removed.
Multi-page crawl (up to 5 pages per domain). AI extracts 23 signals: 15 common framework + 8 digital-marketing-specific. Each signal gets a value, confidence score, and quoted evidence.
All 673 agencies have complete signal profiles. The top 20 by visible-signal count are shown below with full transcripts.
The 10K sample already produced 673 fully-screened agencies, each with a complete signal profile. Running the full 719,358 domains at the same 6.73% eligibility rate would yield approximately 48,412 candidates — each with all 23 signals extracted, scored, and evidence-backed.
Eight digital marketing subverticals classified during Pass 0, confirmed during deep extraction.
15 from the common framework, plus 8 signals designed for the digital marketing vertical.
Each signal returns a value, a confidence score (0–100%), and a quoted evidence snippet from the company’s own website.
Ranked by number of high-confidence signals extracted. These are 20 of the 673 screened agencies from just the 10K sample. Every visible signal shown with its value, confidence, and quoted site evidence. Client deliverables carry full company names, domains, and source URLs.
Total cost for this 10,000-domain sample run across both passes.
10,000 domains classified by GPT-4o-mini at 100 concurrent threads.
673 eligible domains deep-screened with multi-page crawl at 60 concurrent threads.
Digital marketing is one of hundreds of verticals we screen. Send your thesis and get a full-vertical census — every company scored and ranked with full signal transcripts, not a sample.
Request the specimen report