Most operations use a fraction of the identity signal available.

Ask a collections operation who it's collecting from and you'll usually get a thin answer: a name, an ID number, a phone number, a balance, maybe a bureau score. That's not an identity. That's a record.

The reality of a customer is far richer — and most of the richness is knowable. Map it out and the picture radiates from a core identity through more than a dozen distinct signal categories, each with its own sub-signals.

A sample of the taxonomy — the full map spans sixteen categories. Source: Nexalta collections advisory framework.

Why this is a data-science problem

Collections is usually run as an operational problem — dialler strategy, agent scripts, contact rates. Those matter. But underneath them sits a data problem that most operations never treat as one: how well do you actually understand the person on the other end of the call?

Two customers with identical balances and identical bureau scores can sit in completely different realities — one with liquidity and a stable income, one a missed pay-date away from default. Treat them the same and you'll misjudge both: wasted effort on the one who was always going to pay, missed opportunity on the one who needed a different arrangement.

Using the signal you already have

The point isn't to boil the ocean. It's to recognise that a far more complete identity picture is available than the one most floors act on — and that every additional signal used well is a sharper decision about who to contact, when, how, and with what offer.

That's the difference between collecting by volume and collecting by understanding. In the hardest market in the world to collect debt, understanding is the edge.