A company record is not a contact.

A company record gives you the starting point: the organization, its known people and the structure around them. Workloom promotes named people into contactable people, then places each one against the org chart. That matters when the record includes a founder, a head of finance and several managers who look similar in a flat export. The person is not treated as an isolated row. Their role sits inside the company where it belongs.

The contact record can include a verified work address and a mobile number, produced in-house. The enrichment waterfall runs our own collection first and only falls back where it has to. That order is deliberate. Every extra handoff creates another place for a stale title, a mismatched company or an address with no accountable verdict to enter the record. Start with the source that owns the context, then add only what is missing.

This is the point where outbound either becomes targeted or turns into clerical cleanup. If you are building a segment in discovery, use the company record to define the account first, then choose the people who actually fit the buying motion. If the record cannot support that decision, do not hide the gap with another filter. Keep the person tied to the account, expose what is known and let the workflow stop where the evidence stops.

Companies invent titles constantly. One calls the owner of a function a VP, another uses chief, another uses a phrase that sounds senior but carries no authority. Workloom normalizes job titles into a role taxonomy, so filtering by seniority works across those variations. The record still retains the person's stated title, but the operational layer gives it a consistent role and level for selection.

That distinction keeps a campaign from mixing the wrong people under one label. A manager, a director and an executive can all appear under a broad department filter, while a specialist may carry a title that looks more senior than the job itself. The taxonomy gives the filter something stable to work with. The org chart then adds context: who sits above the role, who reports into it and where the person belongs.

Use that structure before writing the sequence. In outreach, a message for a functional owner should not inherit the same premise as a message for an executive sponsor. Pick the role and seniority first. Then inspect the surrounding chart. A named person without that context is still a guess, even when the address is valid. The job is not to collect names. It is to identify the right person for the account.

Workloom's adaptive guesser handles nicknames, around 19 address patterns and per-domain convention learning. It looks at how a company actually forms addresses, rather than assuming every domain follows first name dot last name. A person named Robert may use Rob. A surname may be shortened. A company may use initials or a different separator. The guesser accounts for those cases before presenting an address for verification.

The failure mode is not that a pattern is missing. It is pretending certainty when the domain accepts everything. Some domains return a positive response for any mailbox-shaped address, so a basic check cannot distinguish a real person from a well-formed fiction. The adaptive guesser gives up honestly in that situation. That leaves the record unresolved instead of turning an ambiguous domain into a false green light.

This is useful operationally because uncertainty stays visible before a message enters outreach. A team can review the unresolved contact, choose another person on the chart or exclude the domain from a send. It can also see why the record stopped: not because the system lacked a pattern, but because the domain provided no trustworthy distinction. Honest failure costs less than burning a sending identity on a polished guess.

Three independent checks stand behind every address. Two run in parallel, so the first verdict does not depend on a single path or one moment of availability. A third exists purely to break catch-all verdicts, where a domain appears to accept every address. That separation matters because address syntax, domain behavior and mailbox evidence are different questions. One signal cannot answer all three.

The result is a status with a useful boundary. A verified address has passed the checks that support sending. A catch-all verdict does not get promoted simply because the server answered. An unresolved address remains unresolved. This is the difference between a contact data layer and a list that has been decorated with confidence. When the evidence cannot separate a real mailbox from a plausible one, the record says so.

Freshness is part of the verdict, not a separate housekeeping task. Verified results stay fresh for 30 days and expire at 90, so a send does not rely on a year-old decision. Before a campaign runs, the workflow can work from current statuses instead of assuming the record aged well. The broader controls live in platform, where contact state, account context and outbound execution have to agree before a message moves.