Wiki · Grading
How grading works
What the grade measures, where the grading system stands as of July 2026, and what it takes for your company to hold one — from a first-week provisional self-assessment to an independently struck, board-ready grade. The letters throughout use the same scale the product persists.
1 · Status — July 2026
Where the grading system stands
The foundation is live today: the monthly two-axis grade engine, the API that serves it with tenant and role scoping, the persisted grade snapshot, the letter scale, and the customer-facing grade page. What they produce is a statement of evidence completeness — a pair of measurements about the audit trail, not a performance score, and never a peer ranking. Three gaps remain before that pair is defensible for any company. First, the shipped computation scores each axis by counting qualifying ledger entries, so every entry weighs the same whether it carries five dollars or a million — where the published method defines count coverage over the headline’s volume and rate provenance over its dollars — and the computation must be aligned with those definitions. Second, loading a real company’s inputs — measured usage, cash-ledger evidence, attribution chains, finance-signed rates — is still substantially concierge work, done with you and checked by hand rather than clicked through. Third, the grade records who struck it and when, but the complete independent strike workflow is not built yet. Grades in the product sandbox are seeded demonstrations of the interface and data shape, not production evidence about any real company.
2 · The path to your grade
What it takes to get your company graded
Select the outcome you need. In the first week a company connects measured usage, loads its initial cost and value evidence, and enters provisional rates — enough for a labeled provisional self-assessment. By around day 30 there is enough observed usage, baseline evidence, and finance review to verify a first slice of the ledger. By around day 90 the evidence has accumulated across the period, finance has signed the relevant inputs, exceptions are visible, and an independent reviewer can strike the grade. Those clocks measure evidence accumulating under observation — engineering effort does not compress them, and the remaining product work is tracked separately below.
Product status · July 2026 — select the outcome you need.
Provisional self-assessment
Self-assessed — labeled as suchTypical evidence time · First week
A labeled provisional pair your team can socialize internally, with rates marked pending finance sign-off.
Concierge today — loaded and checked with you, not yet self-serve.
Already shipped
- Two-axis grade engine, monthly snapshot, and letter scale
- Grade API with tenant and role scoping
- Customer-facing grade page
- Metered usage capture and pricing
Your company supplies
- A connected source of measured usage
- Initial cost and value evidence
- Provisional rates — your stated loaded rates, labeled provisional
Oabo automates
- Usage metering and pricing
- The monthly pair and its letters
- The provisional label — the grade says what it is
Finance & reviewer approve
- None yet — provisional means finance has not signed, and the label says so
Remaining product work
- Align the shipped pair computation with the published method — count coverage weighed over the headline’s volume and rate provenance over its dollars, not one equal weight per ledger entry
Evidence time is observation time — usage under watch, baselines forming, a quarter accruing. Product work is engineering time. The two run on different clocks — finishing the product does not shorten the evidence.
Illustrative worked example — not customer data. Change anything.
72%
Moves when more of the volume is metered — collector-observed counts instead of estimates.
15%
Moves when assumed rates are replaced by finance-signed derivations.
Count coverage
C · 72%
Share of the headline volume from metered counts
Rate provenance
F · 15%
Share of the dollarization from finance-signed rates
The weaker axis today is rate provenance — that is where the next hour of evidence work goes.
A single blended grade would read F · 44% — and the mirror image of this ledger would read exactly the same. Two different risk stories, one indistinguishable letter. That is why the axes never combine.
3 · What it measures
It grades the evidence, not the company
The grade answers one question: how complete is the audit trail underneath the headline? It grades the evidence, not the company — a strong quarter with weak evidence grades worse than a modest quarter every line of which can be re-derived. Count coverage asks what share of the headline volume came from metered counts rather than estimates. Rate provenance asks what share of the dollarization rests on finance-signed rates rather than assumed ones. Neither says whether the AI is good; both say whether the numbers about it would survive a challenge.
4 · Two axes
Why two measurements and not one score
A blended letter is information destroyed. A ledger that is fully metered but priced entirely on assumed rates carries a completely different risk than one half metered and fully finance-signed — an average cannot tell them apart, and the module above shows the same refused blend for mirror-image ledgers. So the value-ledger grade is reported as a pair — a letter and a percentage per axis, on the shipped 90/80/70/60 scale. The axes are never averaged into one number in the customer-facing grade, API response, or audit presentation.
5 · Raising it
What moves each axis
Count coverage moves when more of the volume is metered: widen collector coverage, meter the outputs still hand-tallied, retire estimated volumes. Rate provenance moves when assumed rates are replaced with finance-signed derivations — the five-artifact rate card (the GL extract, the denominator definition, the allocation basis, the utilization assumption, and the countersignature), re-signed each budget cycle. The axes move independently, and that is the point: the pair tells you exactly where the next hour of evidence work goes.
6 · The strike
A self-assessment until an independent party reviews it
The grade is a labeled self-assessment until an auditor, an acquirer's diligence team, or a regulator reviews it. Once struck, it carries the reviewer's signature and date, and the product says which it is showing you. The grade is not decoration on the audit trail — it is the audit trail's own statement of how much of itself exists.
See it live: Grade — Related: Method · Wiki index