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

# Progress Score

> How the progress percentage is calculated — and what it means.

The progress score is the headline number Waterline produces for each ticket. It's designed to be honest, stable, and explainable — not an optimistic guess.

***

## How it's calculated

Each acceptance criterion is evaluated independently and classified into one of three states:

| State         | Meaning                             |
| ------------- | ----------------------------------- |
| `SATISFIED`   | Clear, strong code evidence found   |
| `PARTIAL`     | Some evidence found, but incomplete |
| `UNSATISFIED` | No meaningful code evidence found   |

The overall progress percentage is:

```
progress % = (SATISFIED + 0.5 × PARTIAL) / total criteria × 100
```

PARTIAL criteria count at half weight. They signal work in progress — meaningful evidence exists, but not enough to call the criterion done.

***

## Example

A ticket with 4 acceptance criteria:

| Criterion                                  | State       |
| ------------------------------------------ | ----------- |
| User can log in with email/password        | SATISFIED   |
| Session persists across page reloads       | SATISFIED   |
| "Remember me" extends session to 30 days   | PARTIAL     |
| Login rate-limited after 5 failed attempts | UNSATISFIED |

```
progress % = (2 + 0.5 × 1) / 4 = 63%
```

The "Remember me" criterion is in progress — it contributes half weight. A fully unsatisfied criterion contributes nothing.

***

## Why the thresholds are fixed

Waterline uses consistent, fixed thresholds to classify criteria — the final score is not a raw LLM output. In practice, the score is stable: the same codebase and ticket tend to produce the same result. The score can change when:

1. New code is merged
2. The ticket description is edited

That stability makes the score trustworthy. You can track it over time and know that changes reflect real progress in the codebase.

***

## What the score doesn't mean

A 73% score means 73% of the acceptance criteria have clear code evidence. It does **not** mean:

* 73% of the code is written
* 73% of the feature is done by time or effort
* The implemented parts are bug-free or tested

Waterline measures **acceptance criteria coverage in code** — a precise, useful signal, and nothing more.

***

## Uncertainty levels

Every analysis includes an uncertainty level:

| Level    | Meaning                                                                  |
| -------- | ------------------------------------------------------------------------ |
| `LOW`    | Strong, consistent signals — the score is reliable                       |
| `MEDIUM` | Mixed evidence — some criteria may be ambiguous                          |
| `HIGH`   | Sparse evidence — criteria may be underspecified or code not yet written |

High uncertainty is a signal to check the ticket's acceptance criteria or verify the index is up to date.

***

## Improving a low score

* **Review the evidence list** — are the relevant functions showing up? If not, the code may not be indexed yet
* **Check sync status** — if a recent push hasn't been indexed, check the repo status in Settings
* **Write more specific criteria** — vague criteria like "it should work" are hard to match to code. See [Acceptance Criteria](/criteria) for guidance