> ## Documentation Index
> Fetch the complete documentation index at: https://docs.beaconrevenue.io/llms.txt
> Use this file to discover all available pages before exploring further.

# What changes when a stage changes

> A transition is a dated event, not an edit — what it takes for one to be recorded, and which figures move the moment it is.

A stage does not get edited. It moves when a transition event is recorded — a dated fact carrying the stage left, the stage entered, the moment, and how confident Beacon is that it happened.

<Note>
  This page describes a designed model, not shipped behaviour. The mechanism is specified; no surface records transitions from your data yet.
</Note>

## What it takes to record one

A transition is recorded at **70% confidence or above**. Below that it is flagged as low-confidence and routed for review, and **the stage does not move**. A stage that has not moved is a stage Beacon could not confirm — not a stage that stayed put.

Backward moves are ordinary. An account or deal can return to an earlier stage, and that is recorded like any other transition rather than corrected away.

## What moves the moment it is recorded

**The clock restarts.** [Days in stage](/definitions/days-in-stage) counts from entry into the current stage, so it returns to zero at every transition. The days spent in the previous stage are kept.

**The comparison re-anchors.** Pace is judged against a benchmark held per segment *and* stage. A deal entering a new stage is immediately compared against a different distribution — which is why the same 45 days can be unremarkable in one stage and a serious signal in another.

**The stall band recomputes.** Against the median for that segment and stage: under 1.0× is not stalled, 1.0–1.5× is approaching stall, 1.5–2.5× is stalled, and beyond 2.5× is severely stalled.

**The percentile recomputes.** [Deal velocity percentile](/definitions/deal-velocity-percentile) is re-ranked daily against the new stage's distribution. A low number is fast.

**Progression health re-reads pace.** How promptly a record moves between stages is the heaviest single input into the [health score](/definitions/health-score), so a transition changes the score even when nothing else about the account has changed.

## Where the benchmark comes from

From your own closed deals, not from an industry table — a rolling twelve months, recalculated quarterly. A segment-and-stage pairing needs **at least 10 closed deals** before it produces an empirical benchmark and **30** before it is treated as high confidence. Below that, Beacon uses template defaults and labels them as such.

The thresholds sit on the median rather than a higher percentile on purpose: the median is what separates the half of deals moving materially slower than typical from the half that are not. A higher bar would only flag the extremes.

## What does not change

A transition does not change the segment, does not change the stage set, and does not retrospectively alter any figure already sealed for a closed period. Scores move because a stage moved; a stage never moves because a score did.

## Related

<Columns cols={3}>
  <Card title="Days in stage" href="/definitions/days-in-stage">
    The value and its stall bands in full.
  </Card>

  <Card title="Health score" href="/definitions/health-score">
    How progression pace is scored.
  </Card>

  <Card title="Period close and corrections" href="/period-close">
    Why a sealed figure does not move.
  </Card>
</Columns>


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.