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

# Satisfaction

> Scores by agent and by team, and the one rule that keeps a spoken score and a pressed digit from being averaged into a number that means neither.

**Satisfaction** is a 1-to-5 score per call, summarised by agent and by team
over the last 30 days by default — the window somebody means by "how are we
doing".

You see the whole organisation's scores, not only your teams'. Being able to
read every colleague's satisfaction is oversight, and it is the one place the
role deliberately looks wider than a team.

## Two ways a score is collected, never pooled silently

| Method     | What it is                                                      |
| ---------- | --------------------------------------------------------------- |
| **Spoken** | the caller said something and the model read a rating out of it |
| **Keypad** | the caller pressed a digit                                      |

A score read out of *"yeah, pretty good"* is an interpretation. A pressed
digit is the caller's own statement. Averaging the two without saying so
produces a number that looks like a measurement and is partly an opinion, so
the method travels with every score and **the split sits beside the total
everywhere** — on the screen and in the API.

## Reading the summary honestly

Every average is shown with its **count**, because an average of 5.00 from
one call is not the same claim as 4.20 from two hundred.

Beside it is the **distribution** across 1 to 5, and that is the part to
look at. Two averages of 3.0 can mean *everyone is lukewarm* or *half love it
and half hate it*, and those call for opposite responses.

<Note>
  **A score is attributed when it is given**, to the agent and team as they
  were at that moment. It is not resolved through today's roster later. An
  agent moves teams and a team's members change; recalculating attribution
  afterwards would silently rewrite last quarter's numbers for two teams at
  once.
</Note>

## Where scores come from

One score per call, and the **first** answer is the one kept — a caller
re-asked until they change their mind is not a better measurement.

A virtual agent asks for one as part of the conversation. Where it cannot get
a usable answer in words it falls back to the keypad, and a caller who
presses nothing or presses something out of range simply declined. An IVR
flow can ask for one as a step.

Nobody can submit a score through the API. A score anyone can post is not a
measurement.