playbook-tech-on-website

Run two technographic providers in tandem, union and dedupe the domains, then verify each domain against its live homepage using HTML and response header fingerprints. Expect roughly 60% of a provider-sourced list to survive verification, so pull about 1.7x what you need.

Fetch the homepage with a browser user agent at concurrency 2 and match fingerprints over HTML and response headers. Platform hits like Shopify require header evidence such as x-shopid or an oracle endpoint like /products.json to be graded confirmed.

Provider data reflects what a crawler saw at some unstated past point and is often wrong. In the graded test, 4 of 10 companies returned by a provider's own Shopify filter were not on Shopify, and two were not ecommerce at all.

A blocked response is flagged with blocked: true and a blocked_reason, never treated as a negative. Blocked rows are routed to a rendering proxy for rescue, since a bot wall means the page was never read.

Yes, but only on explicit request. The default deliverable is the per-domain verdict; the opt-in clause path uses a locked prompt with gpt-4o-mini plus code-level post-guards, and runs only on rows where the target technology is confirmed.

Do not use it when you want the software a company sells rather than runs, or for mail-infrastructure targeting. It is designed for offers that only make sense to companies running a specific technology on their site.