Where the company data comes from

Every identified visit starts with the network the request came from. That network is described in public registry data maintained by the regional internet registries: who holds an address range, which organisation it is assigned to, which autonomous system routes it. This is the same reference data network operators rely on.

On top of that sits business connectivity information, which distinguishes corporate uplinks and business ISP ranges from consumer broadband and mobile carriers. This layer decides whether a visit can be attributed to an organisation at all — and it is why a home-office visit is reported as an internet provider rather than as an invented company.

A registry entry gives you an organisation, but not a usable profile. The company context you see — legal name, website domain, industry, size band, coarse location — comes from public company information and business directories, matched to the resolved organisation.

The logo cascade is a good illustration of how conservative that pipeline is: it tries provider-supplied branding, then public logo sources, then a favicon, and finally falls back to plain initials. Nothing is invented; when a layer has no answer, the display degrades instead of filling in a guess.

Reference data disagrees sometimes. Ranges get reassigned, companies rename, subsidiaries route through a parent's network, and directories update at different speeds. Where sources conflict, the more specific and more recently confirmed record wins, and where confidence is too low the visit stays unattributed.

Two consequences are worth expecting. A visit from a subsidiary may surface under the parent that owns the network, and a very recent rename may show the previous name until sources catch up. If you see a company card that looks wrong, report it — corrections feed back into the data rather than getting patched only in your view.

There is no cross-site tracking network, no advertising identifier, no data broker profile of individuals and no purchased contact list involved in identification. The output is an organisation, and the inputs are network reference data plus public company information.

The processors involved are listed on the public sub-processor page, which is versioned so you can see what changed and when. If you need contacts inside an identified company, that is a separate step you run through your own enrichment provider under your own contract.