Detect, resolve, and keep it that way.
Four stages, one system: continuous detection, cluster-level resolution with field control, complete record migration, and a reversible audit trail behind all of it.
Matching you can reason about.
Comparison runs per object and per attribute. Exact matching where identity is certain (domain, email, phone) and fuzzy matching where it is not, with normalisation for legal suffixes, punctuation, subdomains and free-mail addresses.
Works across Companies, People, and any custom object with its own attributes.
The surviving record is assembled, not chosen.
A native merge makes you pick a winner and lose the rest. Dedupe works attribute by attribute: keep the oldest created date, the current owner, the longest description, the enriched headcount, from whichever record holds the best value.
Preselected, not decided
Save the decision once
A queue, not a hunt
History follows the record.
The reason manual merges are feared is what they leave behind. Dedupe moves every attached object to the surviving record before the duplicate is retired, and reports exactly what moved.
Nothing is silently dropped. If something cannot be moved, the merge stops and tells you why.
Nothing is destroyed. Everything is logged.
Each resolution writes an entry: the records involved, the winning value per attribute, and where every attached object landed. Rollback restores the prior state exactly, including the records that were retired.
See what is actually in your workspace.
The scan is read-only. You get the cluster count and the resolution plan before anything is written.