Scoring
Every account your organisation owns, scored across 5 domains against the Well-Architected Framework, refreshed daily and decomposable to the individual finding.
Security, cost, reliability, performance and operational excellence. The same method runs against every account you own, so two teams can be compared without an argument about whose reviewer was harsher. Sustainability is deliberately out of scope rather than estimated — a domain we could only infer would make the number less trustworthy, not more complete.
Nothing is secretly weighted. A severity-weighted composite is a judgement call, and the first person to dispute the weighting disputes the whole number. Severity still decides what your team does first — it just does not quietly move the score while nobody is looking.
A score is only useful if someone can ask why. Each domain opens into the questions underneath it, each question into the practices, and each practice into the accounts and resources that failed it. There is no point at which the answer becomes "the model says so."
Scores are kept, so you can show a board or a customer that posture improved over two quarters rather than asserting it. When something regresses, the history says when it last passed and what changed.
Invite-only while we onboard deliberately. We reply either way.