Skip to main content

Understanding Investor Clustering & Confidence Matching

How Recruiterflow clusters investor name variants and scores them against existing company records.

Written by Amogh Balikai

The Investor Mapping Agent does two things before you ever see a result: it groups investor name variants that likely refer to the same investor, then matches each group against your existing company records with a confidence score.

Clustering string variants

AIRA company enrichment writes investor names as free text, so the same investor often appears differently on different company profiles — "Accel Partners" on one, "Accel" on another, "Accel Fund IV" on a third. The agent clusters these variants together so you resolve them once, not three separate times.

Example: Three of your portfolio companies list "Accel Partners," "Accel," and "Accel Fund IV" as investors. Instead of three unrelated entries, the results page shows one group — titled "Accel" — listing all three variants and every portfolio company each one appeared on.

Confidence matching

Each cluster is matched against your existing Recruiterflow company records and scored using the same High / Medium / Low / below-threshold banding as the Duplicate Companies Detection Agent.

  • If one match clears the threshold confidently, it's pre-selected as the suggested profile.

  • If multiple candidate companies are plausible, all of them are shown with a sensible default pre-selected — you choose.

  • If no match clears the threshold, the group shows only Search/Create and Dismiss — there's no suggested match to map to.

How groups are named

A matched group is titled using the existing company profile's name. An unmatched group is titled using its most frequent string variant — the one that appears on the most portfolio companies. Ties are broken by whichever variant was added to a company profile first. This same name pre-fills the new profile name if you choose Create new profile.

Did this answer your question?