You can now see a fairer coverage score on analyses run before depth scoring existed. Earlier reports were held back by a part of the formula they had no data for, so many scores moved up without any change to your content.
The coverage score blends two things: how well your concepts are integrated, and how deeply your page answers eight diagnostic questions. Analyses run before depth scoring existed had no answer for that second part, so the formula counted it as zero. That quietly capped most older reports far below where their actual content deserved to sit.
This surfaced while reviewing why a report could show every essential concept present yet still carry a low score. The two numbers were measuring different things, and the score was being penalised for missing data rather than weak writing.
Depth scoring shipped two days ago, which made the gap obvious: the newest analyses looked healthy while the large back catalogue stayed artificially low through no fault of the content. Leaving it unfixed meant the score read as arbitrary.