Why is my score low when an area only shows 1–3 errors?

[albertfranquesa]

Came up from an early-access user this week and it's a good one. Their Architecture area showed a single issue but scored 57% and a FAIL chip — while another area with far more findings scored higher. If the score isn't counting errors, what's it measuring?

[albertfranquesa]

The score measures rule health, not the number of findings. Here's the model:
Every rule that runs gets scored on its worst finding, not how many times it fired:

No finding → 100 (clean)
Worst finding is LOW → 75
Worst finding is MEDIUM → 50
Worst finding is HIGH → 0 (failed)

An area's sub-score is the average of all its rules' scores, rounded. So one HIGH finding can tank an area even though it's "only one error" — that rule just went to 0, and it drags the average down.
Worked example matching the case above: an area with 7 rules where 1 fires HIGH and the other 6 are clean scores (6×100 + 1×0) / 7 = 86%. If that one rule had fired MEDIUM instead of HIGH, it'd be (6×100 + 1×50) / 7 ≈ 93%. Same single error — very different score, because severity is the lever, not count.
The fix priority that moves your score fastest: open the lowest-scoring area, fix HIGH findings first (0 → 100 is the biggest jump), then MEDIUM, then LOW, then re-scan.