These engine reports are obviously complete utter rubbish. The move by move engine evaluation is a much better gauge of move strength.
They aren't. It's a percentage. If you read into it wrong then that's your own fault. If someone has 90% accuracy in a 50 move game it is telling you it didn't like 10 moves. 10 moves that could be blunders. It's not a damn report card.
It also cracks me up when the report labels what may be the second "best" move as "excellent," even if it is a losing move and/or significantly worse than the "best" move.
And I'm still trying to wrap my head around "brilliancies," and how and/or why they initially fool the engine as opposed to other material sacrifices.