What a RACI matrix is
The matrix is a grid. Each row is a decision or deliverable, each column is a role, and each cell holds one of four letters or stays empty. Its purpose is to make ownership explicit before a deadline tests it. Most marketing work that goes wrong between teams goes wrong at a hand-off or a decision that everyone assumed someone else had, and a RACI exposes those assumptions while there is still time to settle them.
Reading the finished grid shows four kinds of problem at once. A row with no accountable role is a gap. A row with two is an argument waiting to happen. A column full of responsible entries is an overloaded role. And a column full of consulted entries usually means someone is being asked for views out of courtesy, slowing every decision they touch.
This page treats the matrix as a stand-alone tool for any shared piece of work: a launch, a campaign, an approval path, a partnership or a reporting process. Applying it to the shape of a marketing team, alongside choosing between functional, channel and pod structures and setting spans of control, is covered in the guide to marketing team models and organisation design, which uses the matrix as one part of designing the team.
Columns are usually roles rather than named people, so the matrix survives a change of staff. Record the current holder of each role alongside it, because a role nobody currently fills is a common and invisible gap.
The four roles
| Role | Meaning | Rule |
|---|---|---|
| Responsible | Does the work or makes the decision happen | At least one per row; several are possible, but a long list suggests the row needs splitting |
| Accountable | Owns the outcome, signs it off and answers for it | Exactly one per row; the accountable role may also be responsible |
| Consulted | Gives input before the decision or the work is final | As few as the decision needs, each with something specific to add and a deadline for adding it |
| Informed | Is told once the decision is made or the work is done | Told promptly and through an agreed route; being informed carries no veto |
The distinction that causes most confusion is between responsible and accountable. The responsible role does the work, such as drafting the creative, building the campaign or preparing the price list. The accountable role decides when it is good enough, signs it off and answers for the result. On small teams they are often the same person; on larger ones, separating them is the point.
The distinction that matters most for speed is between consulted and informed. A consulted role is a two-way conversation: the work waits for its input. An informed role is one-way: nothing waits. Every consulted entry is therefore a potential delay, and the length of the consulted column largely sets how quickly decisions get made. Some teams use variants that add a role for supporting work or for a separate reviewer; they work the same way, and the single-accountable rule carries across.
Building one around decisions
Start with the decisions that cause arguments, not the tasks. "Set the launch date", "approve the Creative BriefA short document telling the people making creative work what it must achieve, for whom, with which message and within which limits.", "change the Lead RoutingRules that assign each new lead to an owner, queue or next step based on territory, segment, score or source. rules", "approve product claims" and "release spend beyond the agreed budget" are the rows worth writing. "Write the email" rarely causes a dispute, and task rows swell the matrix until nobody reads it.
Then list the roles that touch those decisions, including those outside marketing. Sales, product, finance, legal, regional teams and agencies often hold a consulted or accountable cell, and leaving them out produces a matrix that marketing agrees with and nobody else recognises.
Fill the accountable column first, one role per row. Where two teams both claim a row, that disagreement is the most valuable thing the exercise will surface. Two questions usually settle it: who answers for the outcome if it goes wrong, and can the row be split into parts that each side genuinely controls? A launch date, for instance, may split into the date the product is available, owned by product or buying, and the date it is announced, owned by marketing. If neither question settles it, the person with authority over both teams decides; a compromise that leaves two accountable roles settles nothing.
Then fill responsible, consulted and informed, in that order, and challenge every consulted entry. What would this role add, and what happens if it does not reply in time? A rule that silence after an agreed deadline counts as agreement, or a named route for escalation, stops consultation from turning into a veto.
The guides that cover specific shared work show where the matrix fits. The guide to marketing approvals and compliance reviews decides which reviews a piece of work needs according to its risk; the RACI then records who holds each review. The launch planning and tiers guide names an owner for each launch workstream; a RACI adds the decisions that cross workstreams. And the comparison of central and local marketing control describes a hybrid model whose division of decisions between the centre and its locations is, in practice, a RACI. The tutorial on writing a marketing operations charter uses a short RACI of the decisions that cause disputes as the charter's decision-rights section.
Keeping it alive
A matrix decays quietly. People change roles, teams are reorganised, new channels appear and old decisions stop mattering, and a grid that was accurate at kick-off can be misleading within months. Give the matrix itself an owner, and review it when something changes: a reorganisation, a new person in an accountable role, a new agency, or a dispute that the matrix should have prevented.
Keep it short enough to read in one sitting, which in practice means one page of the decisions that matter. Keep it where the work happens: in the launch plan, the project tool or the brief template, not in a folder nobody opens. When a dispute arises, check the matrix first; if the row was wrong or missing, fix it then.
Where an outcome has a measure, the accountable role should usually own the KPIA metric tied directly to an objective. that tracks it, so that ownership on paper and responsibility in the review meeting point to the same person. The comparison of centralised and decentralised marketing operations shows how shared standards and reporting keep a split of decisions working once it is agreed.
Not every piece of work needs the full grid. A single contested decision is often better served by a short decision record that states who decides, by when, after hearing from whom, and what was decided.
Worked example (illustrative)
An outdoor-gear retailer writes a launch RACI
Mistlecombe Outdoor Gear, an invented retailer with 12 shops and an online store, launched its spring range 9 days late last year because the buying team told suppliers 3 March while the e-commerce team built the campaign for 12 March; each believed it owned the date. This year the marketing director drafts a RACI before kick-off, with 8 rows and 6 roles. The launch date row splits in two: the date stock is available, with the buying director accountable, and the date the range is announced, with the marketing director accountable and buying consulted. Pricing has the buying director accountable and finance consulted. Creative approval has the head of brand accountable, the campaign manager responsible and legal consulted only on claims about waterproof ratings, which cuts the consulted list for that row from 6 roles to 3. Communications to 40 wholesale partners have the wholesale manager accountable and stores informed. Agreeing the matrix takes 2 meetings, 50 minutes of them on the date rows. The range launches on the agreed day, and a pricing dispute in week 2 is settled by checking the pricing row.
Limits and criticisms
It records who does what, not how well or how fast. A matrix with every letter in the right place can sit over a slow process and weak work. Ownership is necessary for good execution, not sufficient, so the matrix needs deadlines, service levels and quality checks beside it.
Matrices grow too long to read. Listing every task produces a spreadsheet that nobody opens and nobody updates, and a stale matrix is worse than none because people trust it. Restricting rows to contested decisions keeps the grid usable.
Long consulted lists slow decisions. Consultation is often granted to avoid offence, and each courtesy entry adds a wait to every decision on that row. Moving roles to informed, and giving consulted roles a deadline, protects the pace.
The single-accountable rule clashes with genuinely shared outcomes. Pipeline depends on both marketing and sales; a brand's reputation depends on the centre and every region. Forcing one accountable role onto such an outcome can misrepresent how it is produced and breed resentment. The usual answer is to split the outcome into the decisions each side controls, give each decision one owner, and measure the shared outcome jointly.
It cannot fix conflicting incentives. If a central team is judged on brand consistency and a regional team on local sales, a RACI will record their dispute without resolving it. The targets and rewards have to be aligned as well, or the matrix becomes a record of who to blame.
Related frameworks
Phase-gate reviews split a launch or market entry into stages with a decision at each gate, and each of those decisions needs one accountable role, which the RACI supplies. The idea of one-way and two-way door decisions helps size the consulted column: a decision that is hard to reverse justifies wider consultation, while an easily reversed one should have a short list and a fast owner.
Common mistakes
- Two accountable roles on a row. Shared sign-off sounds fair and leaves every hard call unmade.
- Rows for tasks instead of decisions. The matrix swells, and the rows that cause arguments get lost.
- Consulting everyone. Each extra consulted role is a wait on every decision in that row.
- Treating informed as consulted. Someone told after the fact who expected to be asked becomes an opponent.
- Roles nobody holds. A column for a role that is vacant or unclear hides a gap behind a tidy grid.
- Writing it once. A matrix not reviewed after a reorganisation describes a team that no longer exists.
- Using it to settle incentive conflicts. When teams are rewarded for different results, fix the targets as well as the grid.
