ODAMReference

The maturity profile: grades and examples

How each of the seven questions is graded, when grades go down, what a review has to record, and two complete worked examples.

Back to the maturity profile

1The seven questions

The seven questions, asked of one outcome
PracticeQuestion
AimWhat must stay true, in what range, and who owns it?
MeasureHow will we know it's holding, and what decision does the number drive?
Evidence neededWhat do we have to see to know?
CoverageDo we actually see it, and is what we see fresh?
DetectionWhen it breaks, do we notice, and has that been tested?
Bounded workWho or what may act on it, within what limits, and who checks?
LoopDo we learn, and are repairs verified rather than assumed?

2Grades, in full

The five grades, plus Unknown
GradeMeaning
AbsentNothing exists.
StatedWritten down, with an owner named.
In placeIt exists and runs.
CheckedA named person other than the one who did the work reviewed it against a stated criterion, with the evidence, the result, the date and any relationship disclosed.
HeldStill true within its horizon, and reconfirmed after the last change.
UnknownNobody can find out yet. A finding, never a zero.

Graded by what you can show, not by what you own. Unknown is a finding, never a zero.

3When grades go down

  • Held lapses to Checked when its horizon passes.
  • A change drops affected answers to In place until someone checks again.

4What a Checked review records

A named person other than the one who did the work, the criterion, the evidence, the result, the date, and any relationship between reviewer and reviewed.

  • A small shop reaches Checked by borrowing a reviewer from a neighboring county, a state program or a university, not by hiring.
  • Checked is an ODAM practice grade. It is not a claim of independent validation, and a reciprocal arrangement is disclosed as one.
  • Checked records that the review happened, not that every test passed.

5Example A: a county shop (illustrative)

Outcome: "Residents can pay their water bill online." A three-person county IT shop, one outcome.

Example A: each question, its grade, and the evidence shown
QuestionGradeEvidence shown
AimHeldWritten aim: at least 99% of online payments post within one business day each month; owner: utility director; horizon: one budget year; last reconfirmed at the March budget review, after the new payment vendor went live
MeasureStated"Share of payments posted within one business day, per month," named but not yet pulled
Evidence neededStatedVendor settlement report; billing import log
CoverageUnknownNobody knows whether import failures are logged. This is the finding.
DetectionAbsentFailures surface when residents call
Bounded workIn placeOnly the clerk reruns the import; nobody reviews it
LoopAbsent"We talk about it when it breaks." Nothing written down and no owner, so not yet Stated

6Example B: a state agency (illustrative)

Outcome: "Unemployment claims are paid on time." A state agency, one of its forty outcomes. Every value is invented to show the method; no real agency, deadline or statute. Record identifiers such as AUD-0912 are examples of what a review cites.

Example B: each question, its grade, and the evidence shown
QuestionGradeEvidence shown
AimHeldWritten aim: 95% of first payments within the program's set deadline; owner: program director; horizon: one quarter; last reconfirmed at the Q3 review after the payment-system change, against the aim's written criteria (record AIM-Q3)
MeasureIn placeA weekly report of claims paid within the deadline runs, with a written definition. No one other than the analyst who built it has reviewed it yet
Evidence neededCheckedAn internal auditor, named in record AUD-0912 with criterion, evidence, result and relationship (same agency, separate reporting line), reviewed the requirement list against the aim on Sept 12. Result: two requirements added
CoverageIn placeThe monthly coverage check runs; 11 of 14 required sources are onboarded. Identity-proofing vendor logs are missing: a gap, not a finding about fraud
DetectionCheckedA security tester who didn't build the detections, named in record TST-Q3-04 with the criteria, evidence and relationship, ran an outage and a fraud-ring scenario last quarter. The outage was detected; one fraud pattern was missed. Checked means the review happened, not that every test passed
Bounded workIn placePayment holds need two people, and automation can recommend but never release. No recorded check of that separation yet
LoopStatedPost-incident reviews are written and owned, but repairs aren't re-measured, so verified repair is missing

The three numbers for the same outcome, Q3 (illustrative)

  • Days at Risk: median 6, P90 41 days, across known exposures on claim-processing systems; clock from the evidenced start; 12 still open, shown separately.
  • Time to Good Decision: P90 3.5 hours for identity incidents, 7-day validation window; 2 reversed and 1 open, shown separately.
  • Ambient Breach: the identity-proofing backlog grew in most weeks of the quarter. A growing backlog alone isn't the Kinetic Trust Protocol's capacity threshold.

Coverage and Loop are the places to investigate the long risk tail and the identity-proofing backlog. The same questions apply as in Example A; scope, authority and resources differ.

7Rules for reading a profile

  • One profile per outcome. Never summed, averaged or turned into a level.
  • Shared questions don't imply identical authority, or comparability across organizations.
  • The three numbers are read beside the profile, never folded into it: Days at Risk with Bounded work and Loop; Time to Good Decision with Detection and Coverage; Ambient Breach with Coverage and Loop.
Back to the maturity profile