Behaviour over time

Behaviour is a shape, not a total. Two customers with identical lifetime spend can be on opposite trajectories, and the total hides which is which.

In Nexyron every subject carries its own sequence of events with time attached: a customer's payments, an asset's faults and services, a site's daily trade. That history is part of the subject, not a separate table joined back to it.

The questions this makes ordinary

Trajectory. Not what a customer is worth, but whether they are climbing or falling, and since when.

Cohorts that mean something. Grouping by when subjects started, or by what they did, rather than only by which category they were filed under. Comparing this year's intake against last year's at the same age is a natural question.

Recency and tenure. How long since the last event, how long the relationship has run, how those two interact. Most early-warning signal lives here.

Rhythm and its breaks. A subject that acted monthly and has now missed two cycles has told you something a total never will.

Periods that align. Comparing months, quarters or weeks without rebuilding the definition each time, and without a partial current period quietly pretending to be a full one.

Why partial periods matter

A month still in progress will always look like a collapse next to completed months. It is the most common way a trend chart lies.

Nexyron knows the range your data actually covers and treats the boundary explicitly, so an answer distinguishes a real decline from a period that has not finished yet.

Where it leads

Once history is part of the subject, forecasting and risk scoring stop being a separate discipline requiring an export. Both are questions asked of the same model, against the state of the business today.