Which matrix fits your situation?

Summary

Thirteen matrices, three families, one question: when several people touch a piece of work, who is actually on the hook for what? Collectively they are role-clarity matrices. They look interchangeable and they are not. Ten of them assign roles to tasks and are all variations on RACI. Two assign roles to decisions, because a task matrix tells you who does the work but not who gets to say yes. One drops the grid entirely and puts a single name against each outcome. This article is the map. It sets out what every matrix in the set shares, then lays their differences side by side, dimension by dimension, so you can see exactly what each one adds, what it costs, and when to reach for it. If you want the short version: start with RACI, add the one letter that fixes your specific gap, and stop there.

The one thing they all share

Every matrix here exists to kill the same failure: diffused responsibility. When a job is spread across people, each person quietly assumes someone else has it, and the work falls through the gap. Psychologists put a name to the underlying effect in 1968: diffusion of responsibility, the same mechanism that keeps bystanders from acting in a crowd. These matrices are the managerial answer to it. They force a team to say, out loud and in writing, who owns each piece before the work starts.

Three commitments run through all thirteen:

• They name people against units of work. The unit changes from family to family, but the shape is constant: a defined piece of work on one axis, named people or roles on the other.
• They insist on a single owner. Whatever the letters, one role per unit carries final accountability. Splitting it defeats the purpose. This is the rule every matrix guards most jealously.
• They separate doing from owning, and owning from being consulted. The whole family exists because those three are routinely confused, and confusing them is how work stalls.

Everything else is variation on those commitments. Understand them first and the differences below read as a set of deliberate trade-offs rather than a confusing alphabet.

The three families

The families differ in one thing only: the unit of analysis. Get that right and the choice between them is almost made for you.

Family 1, responsibility assignment matrices (task-centric). One row per task or deliverable, one column per person. They answer “who is on the hook for getting this done”. RACI is the baseline and every other member is RACI plus or minus a role. Reach for this family when the work is a set of activities to be delivered.

Family 2, decision-rights frameworks (decision-centric). One row per decision, not per task. They answer “where does the authority for this call actually sit”. They exist because responsibility matrices handle decisions poorly: a grid can tell you who builds the thing without ever saying who approves building it. Reach for this family when the friction is about authority, not effort.

Family 3, ownership (single-owner). Not a grid at all. It collapses responsibility onto one named person per outcome. DRI is its only member and the minimalist pole of the whole set. Reach for it when ownership is the entire problem and a matrix would be overhead.

A mature program usually runs more than one at once, as separate views: a decision-rights framework for the go or no-go calls, a responsibility matrix for the delivery tasks that follow, and a single named owner wherever a task keeps falling between people.

The whole set at a glance

This table is the reference spine of the article. Every matrix, grouped by family, with its role set, what it adds over plain RACI, the unit it works on, and where final accountability sits. Read down a column to compare one property across all thirteen; read across a row to see a single matrix in full. Each role is a letter chip, and each name links to that matrix’s full article.

How they differ, dimension by dimension

The matrices diverge along a small number of axes. Once you can name the axis your team is struggling on, the right matrix usually picks itself.

Unit of analysis

This is the deepest divide and the one that decides which family you are in. Task-centric matrices map work: rows are things to be delivered. Decision-centric frameworks map authority: rows are calls to be made. The ownership family maps outcomes to a single person and skips the grid. A team fighting about who does the work needs Family 1. A team fighting about who gets to decide needs Family 2. Using the wrong family here is the single most common mistake, and no amount of extra letters fixes it.

Number of roles, and the cost of each

More roles buy more precision and cost more upkeep. RACI’s four are the floor. Most variants sit at five, adding exactly one role to close one gap. RACI-VS reaches six for regulated control. Every added letter is a cell someone has to fill in, argue over, and keep current, so the honest question for any variant is whether the gap it closes is worth the maintenance it adds. The family rewards restraint: pick the smallest matrix that names your problem.

How the “owner” is defined

All thirteen insist on a single point of final accountability, but they place it differently. In RACI it is the Accountable role. RATSI separates that idea into Authority, splitting the person who can say yes from the person on the hook for delivery. MOCHA hands ownership to the Owner and gives the Manager a coaching role beside it. In the decision family it becomes Decide (RAPID) or Approver (DACI). DRI makes the owner the entire model. If your pain is that “the owner” secretly means two different jobs, RATSI or MOCHA is the fix.

How dissent and veto are handled

This is where RACI is weakest and where several variants earn their place. RACI’s Consulted is a voice with no teeth: consulted people can object and be overruled without trace. PACSI replaces that soft role with Control, a reviewer who can reject the work outright. RAPID makes the same move at decision level with Agree, a role whose sign-off is mandatory, turning a suggestion into a gate. If your reviewers feel ignored, you have a veto problem, and PACSI or RAPID is the answer.

Control, quality, and sign-off gates

Regulated and quality-critical work needs named checkpoints, not goodwill. RACI-VS adds Verify (checked against criteria) and Sign-off (a named formal approver). RACIQ adds a standing Quality role that lives outside the sign-off. PARIS carried Review-required and Sign-off-required decades ago. If an auditor would ask “who checked this, and against what?”, you need one of these rather than plain RACI.

Coordination and momentum

A matrix says who owns work but not who keeps it moving. DRASCI adds a Driver who chases progress without owning the outcome, useful when the Accountable person is too senior for the day-to-day. DACI’s Driver plays the same role for a decision. If work is well assigned but still drifts, you have a momentum gap, not an ownership gap.

Handling deliberate absence

A blank cell is ambiguous: it can mean “not involved” or “we forgot you”. CAIRO adds Omitted to say, on the record, that a person’s absence is a decision. It is the cheapest high-value addition in the set, and it matters most in large or political groups where people read silence as a slight.

The overhead-to-precision spectrum

Lay the set on a single line and it runs from minimum ceremony to maximum control. DRI sits at one end: one name, no grid. RACI is the light standard. The five-letter variants add a point of precision each. RACI-VS and the regulated matrices sit at the heavy end, where the documentation is the point. There is no “best” position on this line. The best matrix is the lightest one that still names your problem, because every unused role is pure overhead.

Provenance and evidence

A reference set is only as trustworthy as the sources behind each entry, and this family is unevenly documented. Some matrices rest on published evidence; most rest on practitioner convention. Treat that as information when you choose.

The lesson: RACI, RASCI, and RAPID rest on evidence, so lean on them when you need to justify the choice. The lettered variants are conventions, which is fine, as long as you adopt them because they fit your gap and not because the acronym looks thorough.

The task-centric family, compared

All ten task matrices are RACI with the dial turned. Group them by the gap they close and the family stops looking like an alphabet.

RACI is the baseline: doing versus owning, four roles, two rules. RASCI is the most common step up, splitting active Support from passive Consulted. CAIRO adds Omitted to make non-involvement explicit. DRASCI adds a Driver for momentum. Those three are the everyday extensions.

Then come the control-minded variants. RACI-VS adds Verify and Sign-off for regulated gates. RACIQ adds a standing Quality role. PACSI rebuilds the review step as Control, a real veto. Reach for these when checking and approval are the point.

Finally the structural variants. RATSI splits authority, responsibility, and task into separate roles, edging toward a decision-rights framework. MOCHA separates a coaching Manager from a single Owner. PARIS is the ancestor, worth reading for lineage more than daily use. If two of these look identical to you, they nearly are; pick the one whose extra letter names your actual problem and ignore the rest.

The decision-centric family: RAPID versus DACI

Both put decision authority in writing and both insist on one final decider, but they aim at different rooms. RAPID is the heavyweight, built by Bain for high-stakes calls in large organizations. Its power is the Agree role, a mandatory sign-off that makes veto rights explicit, and its discipline is that exactly one person holds the D. Use it when a big decision keeps stalling and you need to see, precisely, who can block it.

DACI is the lightweight, born at Intuit and spread by Atlassian, tuned for product teams moving fast. Driver, Approver, Contributors, Informed: one person drives, one approves, contributors advise, and the rest are told. It carries less ceremony than RAPID and is easier to run in a standup. Use it when a team needs clear decision ownership without the weight of a formal governance model. Choose RAPID for stakes and scale, DACI for speed and simplicity.

The ownership pole: DRI

DRI is the whole family compressed to a single idea: one named person is directly responsible for each outcome, period. There is no grid, no letters, no cells to maintain. Popularized at Apple, where “who is the DRI on that?” is a standing question, it is the minimalist extreme, at the opposite end of the spectrum from RACI-VS. Its strength is unmissable accountability; its limit is that it says nothing about who is consulted, who supports, or who decides. Use it when ownership is the entire problem, and layer a real matrix on top only when the surrounding roles start to matter.

Choosing between them

Work down these questions and stop at the first yes.

• Is the fight about who decides, not who does? Use a decision framework. RAPID for high-stakes calls at scale, DACI for fast product decisions.
• Is the whole problem that no one owns the outcome? Name a DRI and stop.
• Otherwise you want a task matrix, so start at RACI and add one letter for your gap: Support hidden inside Consulted, use RASCI. Blank cells causing confusion, use CAIRO. Work drifting for lack of a chaser, use DRASCI. Reviewers with no teeth, use PACSI. Regulated sign-off required, use RACI-VS. Standing QA function, use RACIQ. Authority and delivery held by different people, use RATSI. A manager coaching a separate owner, use MOCHA.

If two matrices both seem to fit, choose the one with fewer roles. The cost of these tools is upkeep, and the lightest matrix that names your problem is always the right one. When unsure, the Matrix Finder narrows all thirteen to a single recommendation from a few questions. And when no named variant fits at all, build your own letter set in the template generator; the Clear RACI guide explains how to keep a custom matrix clear.

What was dropped, and why

Two matrices from the original brief were removed because they fail a basic legitimacy test: neither has a documented origin, an authoritative source, or use beyond search-optimized roundup articles.

• CARS (Communicate, Approve, Responsible, Support). Appears only in aggregator listicles, usually described as “niche”. No named originator, no primary source, no organization of record.
• CLAM (Contributes, Leads, Approves, Monitors). Traceable to a single practitioner’s blog post and one marketing article. No documented origin and no standing use.

Both can be reinstated with a provenance caveat on request. The judgment here was deliberately strict: a reference set is only as trustworthy as its weakest entry.

Trivial variants noted but not given their own article

Several variants are simply RACI with the letters reordered or one obvious role bolted on. They do not carry enough distinct substance for a standalone article, so they are noted inside the relevant articles rather than duplicated:

• ARCI - identical to RACI with the codes reordered (pronounced “arky”). Covered inside the RACI article.
• RASI - RACI with Support replacing Consulted. Covered inside the RASCI article.
• RACI-F - RACI plus Facilitation, a niche Scrum addition. Covered inside the RACI article.
• DCI - Decision maker, Consulted, Informed. A stripped RACI. Covered inside the RACI article.
• ARPA - defined inconsistently across sources (Authority, Responsibility, Process, Accountability in some; Accountable, Responsible, Participant, Advisor in others), which is itself a legitimacy warning. Noted here only.

How to read the set

Each matrix has its own article with the same structure: a summary for readers in a hurry, the letters and what they mean, origin and history, the problem it solves, when to choose it against its siblings, common pitfalls, a worked example, where it sits in the family, and sources. Start with RACI, then read the variant that matches your gap.

Frequently asked questions

Sources

Web sources

• Rogers, P. and Blenko, M. RAPID Decision Making. Bain & Company. https://www.bain.com/insights/rapid-decision-making/
• DACI: A Decision-Making Framework. Atlassian Team Playbook. https://www.atlassian.com/team-playbook/plays/daci
• Using the Directly Responsible Individual (DRI) concept at work. BiteSize Learning. https://www.bitesizelearning.co.uk/resources/directly-responsible-individual-dri-apple
• RACI matrix: history of a managerial invention. Perfony. https://www.perfony.com/en/raci-matrix-history-of-a-managerial-invention/

Academic sources

The foundational literature for the set as a whole, cited throughout the individual articles.

• Korey, G. (1973) ‘LRC: modern technique for functional organization analysis’, Academy of Management Proceedings, 1973(1), pp. 184-189.
• McCann, J.E. and Gilmore, T.N. (1983) ‘Diagnosing organizational decision making through responsibility charting’, Sloan Management Review, 24(2), pp. 3-15.
• Gilmore, T.N. and Kazanjian, R.K. (1989) ‘Clarifying decision making in high-growth ventures: the use of responsibility charting’, Journal of Business Venturing, 4(1), pp. 69-83. https://doi.org/10.1016/0883-9026(89)90034-9
• Jensen, M.C. and Meckling, W.H. (1976) ‘Theory of the firm: managerial behavior, agency costs and ownership structure’, Journal of Financial Economics, 3(4), pp. 305-360. https://doi.org/10.1016/0304-405X(76)90026-X
• Jensen, M.C. and Meckling, W.H. (1992) ‘Specific and general knowledge, and organizational structure’, in Werin, L. and Wijkander, H. (eds.) Contract Economics. Oxford: Blackwell, pp. 251-274.
• Fama, E.F. and Jensen, M.C. (1983) ‘Separation of ownership and control’, Journal of Law and Economics, 26(2), pp. 301-325.
• Lawrence, P.R. and Lorsch, J.W. (1967) ‘Differentiation and integration in complex organizations’, Administrative Science Quarterly, 12(1), pp. 1-47. https://doi.org/10.2307/2391211
• Darley, J.M. and Latané, B. (1968) ‘Bystander intervention in emergencies: diffusion of responsibility’, Journal of Personality and Social Psychology, 8(4), pp. 377-383. https://doi.org/10.1037/h0025589
• Rogers, P. and Blenko, M.W. (2006) ‘Who has the D? How clear decision roles enhance organizational performance’, Harvard Business Review, 84(1), pp. 52-61.
• Project Management Institute (2021) A Guide to the Project Management Body of Knowledge (PMBOK Guide). 7th edn. Newtown Square, PA: PMI.