The Miatz Docs Score
A 0–100 rubric score for your documentation, with a fix queue ordered by impact.
- One number for documentation quality, broken into seven areas you can act on
- A fix queue ranked by how many points each fix returns
- Criteria nobody assessed are left unscored, never counted as failures
The seven areas
| Area | Worth | Asks |
|---|---|---|
| Structure | 20 | Do you cover tutorials, how-tos, reference and explanation — each doing one job? |
| Reference quality | 20 | Can a developer finish an integration without leaving your docs? |
| Deep-linkability | 12.5 | Can a person or a model cite one exact section? |
| Freshness | 12.5 | Is the corpus demonstrably current, or quietly rotting? |
| Search visibility | 10 | Can a search engine still index and rank it properly? |
| Answer-engine readiness | 12.5 | When an AI answers about your product, can it use your words? |
| Agent access | 12.5 | Can an agent read the corpus cheaply and cite it precisely? |
Each criterion is scored 0–3. Expand an area on the report to see its criteria and what full marks look like.
The fix queue
The queue is the deliverable. Every entry shows:
- the criterion that is short, and what a 3 looks like;
- the points it returns if you fix it;
- the specific pages the run flagged, where it could name them.
Work down the list and the number moves. That ordering is the whole point — a dashboard tells you where you stand, a work order tells you what to do on Monday.
Honesty rules
- Unassessed criteria are never counted as failures.
- The coverage figure is always shown next to the score.
- Page-level findings come from your own corpus and are shown as plain text.
The Docs Score measures documentation. It is not an accreditation and makes no claim about anyone's skills.
Frequently asked
How is the score calculated?
Seven areas, each criterion rated 0 to 3. Substance (structure and reference quality) is worth 40 points, trust (deep-linkability and freshness) 25, search visibility 10, and the AI channel (answer-engine readiness and agent access) 25. The total is the weighted result over the criteria that were actually assessed.
Why does the report say only part of the rubric was assessed?
Because a run can only measure what it can see. Anything it could not assess is left unscored rather than treated as a zero — inventing a failure would be as dishonest as inventing a pass. The coverage figure tells you how much of the rubric the number is based on.
What does 'points on the table' mean?
The sum of what every outstanding fix is worth. Each queue item shows its own share, calculated from the weight of its area and how far short it currently falls, so the first item is always the one that moves the number most.
Is this an accreditation or a certification of our documentation?
No. It is a documentation-quality measurement against a published rubric. It says nothing about any person's competence, and it is not awarded, accredited or endorsed by any external body.
Related guides
What miatz reads from your documentation, and what it turns that into.
ReadPoint miatz at the places your product truth already lives, and run them.
ReadThe tracks your corpus compiles into, and the lifecycle every piece moves through.
Read