Your go-to tech blog

Only 3 Things Separate RACI From DARCI, and Only One of Them Changes Who Decides

Ask ten managers what DARCI adds to RACI and most will say “a D for the decider”, as if the new letter were the whole story. It is the smaller half.

The bigger change happens to a letter that was already there: in DARCI, Accountable stops meaning “the person who makes the call” and shrinks to “the person who answers for delivery”. If you add the D column and leave the old meaning of A in place, you get a matrix with two bosses on every row and more meetings than before.

That misunderstanding is why so many teams try DARCI for 2 or 3 months, see no difference, and quietly go back to the spreadsheet they had. The argument below is that the switch only pays off when the rows change too: from tasks to decisions.

Accountable was always doing two jobs

In a classic RACI chart, every row is an activity and every column is a person or role. Each cell holds one of four letters. Responsible does the work. Accountable is the single person who answers for the result and signs it off. Consulted gives input before the work is done, in a two way conversation. Informed hears about the outcome afterwards, one way.

The rule most guides repeat is “exactly 1 A per row”. That rule hides a quiet assumption: the person who answers for a result is also the person who gets to decide how it is reached. For a small task, that holds. The owner of “publish the release notes” can also decide their format without anyone getting hurt.

It breaks on anything with trade-offs. Take “choose the payment provider for the new market”. The engineering lead will live with the integration for 3 years and answers for it shipping on time. The finance director cares about the 1.4 percent versus 2.1 percent fee difference. The head of product owns the market launch. Who is A? Teams usually pick the most senior name, which turns A into a rubber stamp, or they write “A” in 2 cells and break the rule they started with.

DARCI’s answer is to pull the decision out into its own letter. Decider makes the final call on a specific question. Accountable answers for the outcome being delivered. The same person can hold both on simple rows, and that is fine. On the rows that matter, separating them is the point.

The three real differences, one by one

The first difference is the new column. D marks the one person whose call is final once the consultation is over. Not the person who suggests, not the person who approves the budget afterwards. When D is filled in, “who decides?” stops being a question for a meeting.

The second difference is the narrower A, and this is where teams slip. After the switch, Accountable owns delivery: the result arrives, at the agreed quality, by the agreed date, and somebody answers if it does not. A no longer gets a veto on how the result is reached, unless that person is also D on that row. If your A holders still behave as approvers, the matrix now has 2 decision points per row and every disagreement escalates.

The third difference is what a row describes. A RACI row is usually an activity: “write the test plan”, “run the user interviews”. A DARCI row works best as a question with a deadline: “which 2 segments do we launch with, decided by 15 March”. Activities have owners. Decisions have deciders. Mixing the two in one sheet is how a 60 row matrix ends up with 12 people marked D.

A related framework sits close by. Bain’s RAPID (Recommend, Agree, Perform, Input, Decide) also gives the decision its own letter, and adds Agree for people with a formal veto, such as legal or compliance. Atlassian’s team playbook popularised DACI (Driver, Approver, Contributors, Informed), where the Approver plays the decider role and the Driver runs the process. If your company already uses one of those, DARCI adds little. The useful idea in all three is the same: one named person per decision.

Counting the decision: a formula for matrix health

You can audit any responsibility matrix in about 20 minutes with two counts per row and one count per person. The checks below are plain arithmetic, so they work the same in a spreadsheet as on a whiteboard.

For every row r:
  D(r) = number of cells marked D
  A(r) = number of cells marked A
  row passes if D(r) = 1 and A(r) = 1

ambiguity_rate = rows that fail / total rows

For every person p:
  consult_load(p) = number of C cells for p x average hours per consultation
  decision_load(p) = number of D cells for p

An illustrative run. A matrix has 40 rows. 6 rows have no D, 3 rows have two. That is 9 failing rows, and an ambiguity rate of 9 / 40 = 22.5 percent. Roughly one decision in four has no clear owner, and those are usually the rows where the meetings run long.

The person counts are just as revealing. If one product director is C on 24 rows at about 1.5 hours each, that is 36 hours of consultation for one project, close to a full working week. If the same director is also D on 19 of the 40 rows, nothing moves while they are on holiday. A rough rule of thumb many teams use: below 10 percent ambiguity, leave the matrix alone; above 20 percent, fix the D column before anything else.

A product launch that stalled on the A column

This case is illustrative, assembled from patterns that come up often in mid size software companies, not a report on a real firm.

A 14 person team prepares a new pricing page for a SaaS product with 3 plans. Their RACI has 25 rows. The head of product is A on 18 of them, because she owns the launch date. Design is R for the page layout, engineering is R for the billing changes, marketing is R for the copy, and finance is C on anything that touches price.

Two weeks before the launch, the team hits a question nobody wrote down as a row: should the middle plan show a monthly or an annual price first? Marketing wants annual, because it lowers the visible number from $49 to $39. Finance wants monthly, because annual prepayment changes revenue recognition.

The head of product is A on the page, so everyone waits for her. She is A, but she does not feel she is the right person to judge revenue recognition, so she asks finance for a written view. Finance replies in 4 working days. The launch slips by a week.

Their DARCI rework took one session. They added 7 decision rows to the existing activity rows, each phrased as a question with a date. On “price display on the pricing page”, the head of finance became D, marketing and design were C, and the head of product stayed A for the page shipping on time. The next pricing question, about a discount banner, was settled in 2 days. The matrix did not make anyone smarter; it made it obvious whose phone to call.

Converting a RACI to DARCI in an afternoon

The conversion works best as one working session of 2 to 3 hours with the people who are named in the matrix, not as a document someone fills in alone.

  1. List the decisions separately from the activities. Go through the project plan and write every open question that has more than one reasonable answer. Most projects of 3 to 6 months produce 8 to 15 of them.
  2. Phrase each decision as a question with a date. “Choose analytics tool” becomes “Which analytics tool do we sign by 30 April?” A date turns a vague topic into something a D can close.
  3. Assign exactly one D per decision. Pick the person who carries most of the consequences and has the information to judge them, which is often not the most senior person in the room.
  4. Re-read every A cell with the narrower meaning. Ask each A holder: “Do you answer for delivery here, or did you expect to approve the approach?” Where the answer is “approve”, either make them D on the related decision row or move them to C.
  5. Run the formula from the previous section and publish the result. Share the ambiguity rate and the 3 heaviest consult loads with the team, then revisit the sheet once a month or whenever a decision gets stuck for more than a week.

Step 4 is the one people skip, and skipping it is how a team ends up with DARCI on paper and RACI in practice.

How to answer RACI vs DARCI in an interview

The question comes up in project manager, product manager and engineering manager interviews, usually framed as “how do you make sure decisions get made on a cross functional team?” A strong answer has three parts and takes under 2 minutes.

Start with the mechanism. RACI ties decision rights to Accountable, which works on simple rows and breaks on trade-off decisions. DARCI separates the Decider from the person accountable for delivery, so the person answering for the date is not forced to judge every expert question.

Then show judgement about when not to use it. For a team of 5 working on one product, a full matrix is overhead; a short decision log does the job. DARCI earns its keep when 3 or more functions share a project and decisions keep coming back to meetings.

Finish with a concrete example, ideally your own. “On my last launch, 4 of our 12 open decisions had no owner, so I added a D column and the average time to close a decision dropped from about 9 days to 3.” Only use numbers you actually measured. Interviewers ask follow up questions on figures like that, and a made up statistic falls apart on the second one.

If you want to try the idea on your current project, take the one decision that has been open longest, write it as a question with a date, and ask your manager to name its D in the next 1 on 1.