DataVisuals™ The decision governance company Start free Score your institution

The Framework

The Decision Governance Maturity Model

A framework for assessing the maturity of a financial institution’s decision governance — its discipline around making consequential decisions. Five levels across six facets. Free to use and cite.

Read the methodology paper →

Section 01 — The five levels

The universal scaffold.

The levels are cumulative — Level 4 requires Level 3, Level 5 requires Level 4. An institution does not skip levels.

L1

Initial

No defined practice. Whatever happens, happens. Nothing to point to, nothing to measure.

L2

Managed

The practice exists in some workflows but not institution-wide. Discipline is local, not systemic.

L3

Defined

A written standard applies across workflows. The standard exists as policy, whether or not consistently enforced.

L4

Measured

Performance against the standard is measured and reportable on demand. Discipline is observable, not just declared.

L5

Optimizing

The standard adjusts based on what measurement reveals. The standard is not static; discipline is self-correcting over time.

Section 02 — The six facets

What the Maturity Model measures.

Six facets — Ownership, Timeliness, Reconstructability, Override Discipline, Authority, and Apex Linkage. They are the six questions a CRO asks reviewing a workflow, a board chair asks reading a risk report, and an examiner asks opening a regulatory file — The Plain Frame. Each facet is scored L1–L5; the institution’s overall placement is a floor-with-exceptions composite across the six.

01 · Ownership

“Who’s doing it?”

Every consequential decision has a named owner on the record.

See the five levels
L1 — Initial
Asked who owns a randomly selected decision from last quarter, the institution cannot produce a name without an investigation.
L2 — Managed
Owner-naming exists in pockets — BSA, credit committee — but not across operations, vendor management, IT, or customer escalation.
L3 — Defined
A written standard for who owns consequential decisions applies across workflows. The standard specifies what counts as consequential and how the owner is recorded.
L4 — Measured
The institution produces numbers against the standard. “96% of consequential decisions in Q3 had a named owner at the time of decision; the 4% gap was concentrated in operations and vendor decisions.”
L5 — Optimizing
The standard is updated based on where measurement reveals it is failing. Workflows the policy did not anticipate become workstreams that sharpen the standard.
Read the full level definition →

02 · Timeliness

“When are they doing it?”

Decisions close inside the cycle the work demands.

See the five levels
L1 — Initial
No cycle time measurement on consequential decisions. No defined start, no defined end, no infrastructure for measurement.
L2 — Managed
Cycle time is measured where regulators require it (BSA alert disposition, for example). Elsewhere it is absent.
L3 — Defined
A written standard specifies how cycle time is measured: what starts the cycle, what ends it, how timestamps are recorded, what SLAs apply by decision class.
L4 — Measured
The institution produces numbers on demand. “Median cycle time on consequential credit decisions in Q3 was 14 business hours; the 90th percentile was 47 hours, up 12 from Q2.”
L5 — Optimizing
When measurement reveals drift, the institution finds the structural cause and adjusts. SLAs themselves are revised as the institution learns where they were uncalibrated.
Read the full level definition →

03 · Reconstructability

“Can they prove they did it?”

Rationale and evidence captured as the decision is made.

See the five levels
L1 — Initial
No defined standard for what constitutes a complete decision record. Reconstructing a record requires assembling it from multiple sources.
L2 — Managed
Completeness practice exists in workflows with strong regulatory documentation requirements (SARs, board minutes). Other workflows lack it.
L3 — Defined
A written standard specifies what a complete decision record must contain. The specifics vary by workflow, but the institution-wide framework exists as policy.
L4 — Measured
The institution samples past decisions and reports against the completeness standard. “In Q3, 87% of sampled decisions met the completeness standard.”
L5 — Optimizing
The standard or the workflow is updated based on where records fall short. The recording mechanism changes, or the standard is corrected where it was over-specified.
Read the full level definition →

04 · Override Discipline

“Should it be overruled?”

Overrides are owned, justified, and pattern-visible.

See the five levels
L1 — Initial
The institution does not recognize overrides as a distinct class of decisions. Departures from defined dispositions happen, but no record practice or aggregation.
L2 — Managed
Some workflows recognize overrides — model overrides, credit committee exceptions — because the workflow’s structure forces it. Elsewhere, overrides are treated as ordinary operational events.
L3 — Defined
A written standard defines what an override is, how it is recorded, and what must accompany it. Applies wherever defined dispositions exist.
L4 — Measured
The institution measures override rates and outcomes. “In Q3, the override rate on the credit model was 8.3%; of overrides where the outcome window has elapsed, 73% performed in line with the override decision.”
L5 — Optimizing
The institution adjusts its underlying disposition mechanisms — refines the model, updates the policy, revisits procedural defaults — based on where overrides are patterning.
Read the full level definition →

05 · Authority

“Who has decision rights?”

Decisions sit at the level the policy says they sit.

See the five levels
L1 — Initial
No documented delegation matrix defining approval authority by decision type. Some authority is well-understood (board, SAR filings) but institution-wide practice is informal.
L2 — Managed
Delegation matrices exist where operations require them — credit authority, legal signing authority. Other workflows operate on informal structures.
L3 — Defined
A documented, institution-wide delegation matrix defines approval authority by decision type and threshold. Incorporates the relevant regulatory anchors (Reg O, SR 26-2, ECOA, BSA/AML).
L4 — Measured
The institution samples past decisions and verifies them against the matrix. Both failure modes — authority shortfall and authority excess — are visible.
L5 — Optimizing
The matrix is refined based on what measurement reveals about how it is actually being applied. Thresholds are recalibrated, roles are clarified, committee scope is right-sized.
Read the full level definition →

06 · Apex Linkage

“Does the board see what produced this?”

Board decisions trace to the operational evidence that produced them.

See the five levels
L1 — Initial
A strategic plan exists, but it does not roll down to operational decisions in actionable form, and operational performance does not roll up to inform strategy.
L2 — Managed
Some strategic objectives connect to operational workflows. The linkage exists in pockets, not as a property of the institution’s strategy-to-operations practice as a whole.
L3 — Defined
A defined practice translates strategy into SMART goals that roll down to workforce, vendor, and operational objectives.
L4 — Measured
The institution measures performance against strategic goals at a meaningful cadence — typically quarterly. The underlying initiatives show measurable effect on the trend.
L5 — Optimizing
The institution adjusts strategy based on what the measurement reveals. Requires L4+ on every other facet as a structural prerequisite.
Read the full level definition →

Section 03 — Corroboration

The enforcement record shows the same gaps.

Separately, we score the public failures. Since September 2025 we’ve read every federal enforcement action from the Fed, FDIC, NCUA, FinCEN, and OCC and scored each for decision-record quality — whether the record showed who decided, on what basis, with what authority. The failures land on the same facets the Maturity Model measures. See the tracker →

The same six gaps

Ownership no one can name, rationale never recorded, authority exceeded, overrides untracked. Independent, public failures — and they cluster exactly where the model says the discipline breaks.

Corroboration, not origin

The framework stands on its own: five levels across six facets, defined in the methodology paper. The enforcement record is a running audit of those facets — evidence that they matter, not the source of them.

Section 04 — From framework to operating system

The Maturity Model is the rubric. The platform closes the gap.

Your Governance Profile shows where you stand against the Maturity Model. The Governance Roadmap promotes each gap into an owned initiative with milestones. Those milestones decompose into decisions that flow through your Inbox alongside operational work. The platform governs its own improvement — the recursion the methodology paper calls Apex Linkage.

  1. Profile Private placement against the Maturity Model. Six facets, floor-with-exceptions composite.
  2. Roadmap Each gap promoted to an owned initiative with milestones. Six facet templates, 125+ pre-built milestones.
  3. Inbox Milestones decompose into governed decisions, tagged Governance, alongside operational work.

Section 05 — Definitions

DGMM & Governance Profile.

Two terms, often conflated: one framework, and your institution’s private placement on it.

Card 01 · The framework

DGMM

The rubric.

Five maturity levels across six governance facets. Published in the methodology paper, citable and versioned.

Card 02 · Your private placement

Governance Profile

Your institution’s placement.

Your confidential placement against the DGMM, computed inside the platform from your self-assessment and decision-record telemetry.

An illustration

Each facet, placed on the same five-level scale.

The Maturity Model places every facet on the same L1–L5 scaffold. The figure below illustrates how a set of placements across the six facets can be read at a glance.

Sample Illustrative placement distribution across six DGMM facets Stacked horizontal bars showing illustrative L1 through L5 distributions for Ownership, Authority, Timeliness, Reconstructability, Override Discipline, and Apex Linkage. Ownership Authority Timeliness Reconstructability Override Discipline Apex Linkage 0% 50% 100%
Illustrative — not real data.

What this is not

  • Not a regulatory rating. The assessment is a read on your own discipline, not a substitute for examination findings.
  • Not a vendor scorecard. You’re placed on your own decision-governance discipline, not on the platform you use.
  • Not a peer comparison. Your placement comes from your own answers; it isn’t ranked against other institutions.
  • Not a measure of decision quality. It measures the discipline of how decisions are made, not whether any one call was correct.

Decision Governance as a category

Decision governance — the discipline of making consequential institutional decisions accountably, observably, and improvably — is an emerging governance category in financial services. It sits alongside risk management, compliance, data governance, and operational risk as a distinct discipline with its own practices, vocabulary, and standards.

DataVisuals authored the first formal framework for the discipline. The Decision Governance Maturity Model, the self-assessment, and the Decision Governance™ platform are the formalization of the discipline as a named, software-supported practice. The trademark protects our product name, not the category — decision governance belongs to the practitioner community that develops, refines, and uses it.

Section 06 — Take your placement

See where you stand.

Eight minutes. Six facets. Your institution’s private placement against the Maturity Model.

Take the Maturity Model Self-Assessment Read the methodology →