Accountability Matrix¶
Template — instantiates Accountability Chain Design
Maps decisions, roles, accountable owners, contributors, consulted parties, and informed parties so responsibility is visible before action occurs.
An Accountability Matrix is a grid that fixes ownership in advance: it lists the recurring decisions or action types down one axis, the roles across the other, and in each cell records who is accountable for the outcome, who is merely responsible for doing the work, who must be consulted, and who is only informed. Its defining move is separating the single answerable owner from everyone else who touches a decision — and doing so before anything happens, so that when the moment comes there is no scramble to work out whose call it was. It is a static map of who-owns-what across many decisions, not a live procedure, a forum, or a repair tracker; a filled cell is a standing claim about ownership, and the rest of the accountability chain hangs off it.
Example¶
A regional credit union has been burned twice by loan decisions that "everybody sort of touched and nobody owned." Before rolling out a new small-business lending product, the risk office builds an Accountability Matrix. Down the side: approve loans under $50k, approve loans $50k–$250k, set the pricing model, grant a policy exception, report a suspected fraud. Across the top: branch officer, regional credit manager, chief credit officer, compliance, the board risk committee. In each cell they place exactly one A and as many R/C/I marks as the decision needs — a branch officer is Accountable for a sub-$50k approval and only Consulted on pricing; the chief credit officer is Accountable for exceptions and Informed on routine approvals.
The finished grid is posted on the lending intranet where every officer can see it. Three months later an examiner asks who authorized a borderline exception. Instead of a meeting, the answer is one cell: the chief credit officer, by name of office, with compliance recorded as Consulted. The matrix did not make that decision well — but it made the owner impossible to lose.
How it works¶
The grid encodes one non-negotiable rule and several conventions:
- Exactly one accountable owner per decision. Every row (or column, depending on layout) carries a single A. Two A's mean nobody owns it; zero A's mean the decision falls into the gap between roles. Enforcing single-A is the whole discipline.
- Distinguish accountable from responsible. The people who do the work (R) are not the person who answers for the result (A). Collapsing the two is the most common way a matrix lies.
- Consulted vs informed. Consulted parties have input before the decision; informed parties are told after. Keeping them separate prevents both bottlenecks and surprises.
- Publish it. The grid earns its keep only when it is visible to the people it governs, so ownership can be checked at a glance rather than reconstructed after a failure.
Tuning parameters¶
- Granularity — one row per decision type versus one per individual decision. Fine granularity makes ownership precise but the grid sprawls and rots; coarse granularity stays maintainable but hides edge cases.
- Owner specificity — name a role/office versus a named person. Roles survive turnover; named people are more inspectable but must be re-pointed constantly.
- Single-A enforcement — strict (a validator rejects any row without exactly one A) versus advisory. Strictness is what keeps the matrix honest; laxity lets accountability quietly diffuse back.
- Visibility scope — who may see the grid. Wide visibility maximizes answerability; narrow visibility protects sensitive structures but weakens the point.
- Refresh cadence — how often the matrix is re-checked against how work actually flows. Stale grids describe an org that no longer exists.
When it helps, and when it misleads¶
Its strength is pre-empting responsibility diffusion: it converts "who owns this?" from a post-failure investigation into a lookup, and it forecloses the classic everyone-was-involved-so-no-one-answers outcome by forcing a single A into every row before action.
Its central failure mode is that a matrix is a map, not a mechanism. A cell filled with a name is a claim of ownership, not the machinery that makes ownership real — there is no record of why a decision was made, no forum that can call the owner to account, and no path to repair. Detached from the rest of the chain it becomes RACI theater: a tidy grid that satisfies an auditor while nothing behind it actually holds.[n1] The other misuse is over-granularity — a matrix so detailed it is never updated, so it certifies an ownership structure that lapsed months ago. The guarding discipline is to keep the grid coarse enough to maintain, enforce single-A mechanically, and wire each owner into the records, forum, and repair path that turn a named A into an answerable one.
How it implements the components¶
action_or_decision— the rows (or columns) enumerate the bounded decisions and action types the chain will govern; the matrix is where the accountable object is first named.accountable_owner— the single A in each cell names who must answer for that decision's outcome.responsibility_scope— the A/R/C/I distinction bounds what each role does and does not own, separating the answerable owner from doers, advisors, and bystanders.accountability_visibility— publishing the grid makes ownership inspectable before action rather than reconstructed after failure.
It records who owns each decision but preserves no reasoning behind them (rationale_record), convenes no forum where owners are questioned (answerability_forum — that is the Answerability Review Meeting), and tracks no repair (repair_or_consequence_path, closure_audit — the Corrective Action Register). Its nearest twin is the Incident Ownership Protocol, which also assigns owner and scope but does so in real time during a live failure and adds escalation_trigger; the matrix is the standing pre-action map, not the live assignment.
Related¶
- Instantiates: Accountability Chain Design — the matrix supplies the owner-and-scope layer the rest of the chain hangs from.
- Sibling mechanisms: Answerability Review Meeting · Corrective Action Register · Escalation Record · Incident Ownership Protocol · Postmortem Action Tracking · Decision Log · Role Charter
Editorial Notes¶
Form Classification¶
Form family: Representation, Specification & Plan
Rationale: The mechanism maps decisions, roles, accountable owners, contributors, consulted parties, and informed parties so responsibility is visible before action occurs, so its operative form is a static or prospective information artifact.
Independent corroboration: The frozen evidence defines Accountability Matrix as 'Maps decisions, roles, accountable owners, contributors, consulted parties, and informed parties so responsibility is visible before action occurs', so its operative form is Representation, Specification & Plan.
Review outcome: Independent reviewer agreement; high confidence.
Origin Attribution¶
Primary origin: Organizational & Management Science
Origin pattern: Single lineage
Present-day reach: Multi-domain
Rationale: The RACI-style separation of accountable, responsible, consulted, and informed roles is a canonical organizational and project-management artifact.
Related originating lineages:
- Engineering & Design — Systems and project engineering widely use responsibility-assignment matrices for complex delivery.
- Public Administration & Policy — Public-sector governance applies the same matrix to decision authority and administrative accountability.
Review resolution: The matrix is a direct organizational responsibility-allocation tool, with public administration and engineering as important institutional uses. It transfers across many domains but is not universal to all mechanism classes.
Review outcome: Reconciled after independent review; high confidence.
Notes¶
The matrix maps many decisions to owners across a set of roles; a Role Charter does the opposite, constituting one office's authority and legitimacy in depth. They are complementary layers — the charter says what an office may do and by whose sanction; the matrix says which of the org's decisions that office actually owns. Do not let a matrix stand in for either the reasoning behind a decision or the machinery to answer for it: an A in a cell is where accountability starts, not where it lives.
[n1] RACI (Responsible, Accountable, Consulted, Informed) is the established responsibility-assignment convention this template formalizes; its well-known pathology is that a completed chart is mistaken for functioning accountability, when a role marked "Accountable" with no records, forum, or repair path behind it answers for nothing. ↩