Catch and merge duplicate contacts

A spa books a walk-in client under "Priya" with one phone number, and two months later the same client calls to rebook and gets saved again as "Priya S" with a slightly different number format — now two reps could plausibly call the same client about two different things, each unaware the other record exists, and the spa's own history of that client's preferences is split across two files that never talk to each other.

Duplicate detection checks new and existing contacts against name, phone, and email to flag likely matches like this before they cause exactly that kind of confusion. A rep or manager reviews any flagged pair and merges them, keeping the combined activity history from both so nothing about that client's past conversations gets lost in the process.

Duplicate contacts across channels are consistently the most common data-quality complaint in any CRM, and the reason is almost always formatting, not a genuinely new person — a number saved with a country code once and without it another time, a name typed with a middle initial in one entry and not another. Indian phone numbers carry their own normalisation challenge: a number saved with a leading zero, without a country code, or with inconsistent spacing can look different enough to slip past a naive comparison even though it's clearly the same number to a human, which is why matching needs to normalise formatting first.

The risk on the other side — a false-positive merge — is arguably worse than missing a duplicate, because it actively combines two people's histories into one incorrect record. Two clients sharing a common first name and similar phone formatting could be flagged as a likely match, and if a reviewer merges them without checking carefully, one client's booking history gets permanently mixed with a stranger's. This is why merging stays a human decision, not automatic on a high-confidence score alone — the cost of a wrong merge outweighs the inconvenience of thirty seconds confirming an obvious match.

Because a wrong merge is a real possibility, however carefully the review step is designed, the ability to undo one within a limited window matters as a genuine safety net. A spa manager who merges two records late on a busy day and only later realises they were different clients needs a way to separate them before the mistake becomes permanent.