Each potential duplicate the Duplicate Companies Detection Agent finds gets a confidence score — a measure of how likely it is that two company records refer to the same organization. This isn't a word-for-word name match; it's a smart similarity comparison, so "Google" and "Google Inc." are recognized as likely duplicates even though the text differs.
The confidence bands
Every duplicate row is scored individually against its parent company and grouped into one of three visible bands:
High (green): greater than 98% match
Medium (yellow): 93–98% match
Low (red): 90–92% match
Anything scoring below 90% isn't shown at all — Recruiterflow doesn't surface matches it isn't reasonably confident about.
Scoring is per pair, not per cluster
A single cluster can contain both High and Low confidence rows, because each duplicate is scored individually against the parent — not against the cluster as a whole.
Example
A cluster's parent company is "Acme Corp." Underneath it, "Acme Corporation" scores 99% (High) and merges safely in one click. "Acme Co — West Region" scores 91% (Low) — worth a quick look before merging, since it might be a genuinely separate regional entity rather than a duplicate.
How results are sorted
Clusters are sorted High confidence first, then Medium, then Low. Within each band, clusters with the most attached contacts and jobs combined are shown first, so you tackle your highest-impact cleanup first.
Things to know
You can merge, dismiss, or reassign the parent on any confidence band — High, Medium, and Low are all actionable, not just informational.
Dismissing a pair tells Recruiterflow it isn't a duplicate, so that exact pair won't resurface in future scans. See the FAQ for how to reverse this.