> ## 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.

# Usage pattern

> The shape an account's adoption has taken over six months, the eight patterns it is classified into, the ninth value that means Beacon is not sure, and the two pairs that are easy to confuse.

## What it is

Measured. The shape an account's adoption has taken over the last six months, classified into one of eight named patterns.

It exists because a usage score on its own is ambiguous. Two accounts can sit at the same number with one climbing and one falling away, and two accounts on the same trajectory can be at completely different depths. **The score says how strong engagement is. The pattern says what shape the curve is.** You need both, and this is the second one.

The pattern is what routes an account to the right intervention. Reading a slow, steady climb as a stall produces an escalation nobody needed. Reading a genuine stall as a slow climb misses the point where intervening would have worked.

## How it is calculated

Measured, and **not a score** — it is a classification. Beacon reads the shape of five signals over the trailing six months and matches it against eight patterns: how activation milestones were reached against the timing typical for the account's segment, how deeply features were adopted, how intensity moved, whether more than one team came on, and how continuous engagement has been.

| Pattern                       | The shape it describes                                                                                                         |
| ----------------------------- | ------------------------------------------------------------------------------------------------------------------------------ |
| `foundation_and_build`        | Strong activation in months 1 to 3, broad adoption from 3 to 6, expanding across teams from 6 onward                           |
| `quick_activate_then_plateau` | Strong activation in months 1 to 2, reaching the typical count, then flat — breadth never expands                              |
| `slow_burn_steady`            | Slower than typical early on, but consistent and in sequence, deepening from months 6 to 12                                    |
| `stalled_onboarding`          | Slower than typical in months 1 to 3 **and** no progression into month 4, with foundational milestones still incomplete        |
| `bounce_and_decline`          | Reaches typical activation in months 1 to 2, then declines across depth, intensity and recency from month 3                    |
| `late_bloomer`                | Slower than typical through month 4, then accelerates from month 5 as fit is found, catching up on milestones by months 6 to 9 |
| `multi_team_expanding`        | Strong single-team activation through month 6, then meaningful adoption by other teams from 6 to 12                            |
| `power_user_concentrated`     | Strong single-team activation that gets more intense but never broadens — high intensity, small base                           |
| `mixed`                       | Beacon is not confident enough to call it                                                                                      |

**`mixed` is not a ninth shape.** It appears when the best match is under 60% confidence and the runner-up is within 20 points of it — meaning two patterns fit almost equally well. It routes to a person rather than to an automated play, which is the correct outcome when the shape is genuinely ambiguous.

**Three of these pairs are the ones worth getting right.** `slow_burn_steady` and `stalled_onboarding` look similar early and mean opposite things. So do `late_bloomer` and `stalled_onboarding` past month 4. And `quick_activate_then_plateau` looks healthy for two months before it stops being so.

`stalled_onboarding` is the only pattern with a hard rule behind it: activation below 40 **and** month 4 reached. That rule cannot be loosened without joint sign-off, because it is the point where intervening still works.

**The eight patterns are fixed.** They are the same eight for every company, which is what lets a pattern mean the same thing in two places. Changing the set is deliberate and takes joint approval across customer success, product and Beacon.

## Where it comes from

| Source                                                                           | What it supplies                                                                                           | Required |
| -------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------------------- | -------- |
| Your product analytics — Mixpanel, Amplitude, Heap, Pendo, or your own event log | Events, feature adoption, active user counts, session distribution, time since last meaningful interaction | Yes      |
| Your segment definitions                                                         | The typical milestone count and timing the shape is judged against                                         | Yes      |
| Your customer success platform                                                   | Context on the account, for display                                                                        | No       |

Beacon reads product usage either through an analytics platform already connected, or through a pipeline built from your product database directly. Either works.

**Nothing is published until there are at least seven days of usage data and the daily coverage is above 95%.** Below either, the signals do not compute — you get an absence with a reason, not a low number.

**One measure has a dependency worth knowing about before you start.** The breadth signal — whether more than one team is active — depends on group analytics, which is a paid add-on on every major platform *and* has to be instrumented in your product before it produces anything. Where it is not in place, **Beacon marks that layer unavailable rather than scoring it zero.** Zero is a measurement; absent is not, and treating one as the other would understate the account with nothing on the record to say so.

## How fresh it is

Classified daily. **But a changed pattern is not declared until it has held for at least fourteen days** — the shape of six months should not flip on one quiet week. The value you see is fresh every day and deliberately slow to move.

Some events are read immediately rather than waiting: an activation milestone reached, a new team becoming active, a feature being retired, or usage dropping by half or more month on month.

Every response carries two timestamps. **`as_of` is the moment the value describes. `computed_at` is when Beacon last worked it out.** Read both rather than assuming a cadence.

## Currency and rounding

Neither applies. It is a category, not a number, and carries no money value.

Enumerated values are lowercase with underscores on every door.

## What changes it

A change in the shape of the last six months — milestones reached or missed against segment timing, features adopted or abandoned, intensity moving, a new team crossing into active use, a gap opening in engagement. Reaching month 4 with activation still below 40 declares a stall.

And a quarterly recalibration of what is typical for a segment, which moves the baseline the shape is read against. Those typical counts start as defaults and are replaced with figures from your own closed accounts once there are at least fifty of them per segment.

What does not change it: anything lasting under fourteen days, and a change in the usage score on its own — the two are built to be independent.

## What it is not

* **Not the usage score.** That is a 0 to 100 measure of how strong engagement is, judged against the account's segment. This is the shape. An account can be strong and falling, or weak and climbing, and only reading both tells you which.
* **Not one of the five signals.** Those are the parts, and each has its own score. The pattern reads how they moved over time, not where they are now.
* **Not a churn prediction.** A pattern is a description of what has happened. What to do about it is a separate judgement Beacon makes elsewhere.
* **Not a three-way trend.** Beacon does not publish a rising, flat or declining direction. There are eight shapes and they are not reducible to three without losing the distinctions that make them useful.
* **Not a count of people.** Beacon publishes how many users and teams were active as counts, never as lists. There is no person-level view of your customers' users at any point.