Most RACIs are paperwork. The strongest are forensic instruments. This one is built to surface accountability problems in time to fix them, not in the post-mortem.
Rows are decisions, not workstreams. Each row has exactly one Accountable (one human, not a team), the Responsibles who do the work, a deliberate Consulted list, and the Informed roster. The format forces the question that prevents two-VPs-disagreeing-quietly incidents.
Who decides? What happens if the decision needs to be made on Tuesday and the Accountable is on vacation? The cheat sheet covers how to use the test to spot the rows that will fail before the program does.
Most PMs let the Consulted list balloon to twenty-three people. The result is decisions that never get made. The guide covers how to spot a C list that is really an I list, and how to cut it without political fallout.
The four letters, the two most common failure modes, the script for telling someone they have been moved from C to I, and the escalation rules for when an A goes on vacation.
One purchase, yours to keep. Use it on every program, customize for your team. Works in Excel and Google Sheets.
I once watched two VPs make the same decision differently in the same meeting. One said the launch date would be the first of the quarter. The other, twenty minutes later, said the date would be the middle of the quarter. Neither contradicted the other. Neither noticed. By the time the discrepancy surfaced a month later, two teams had built work products toward different launches.
The PM had everything she needed to prevent that incident. She just didn't have a tool that forced the question of who actually decided what.
"RACI is a forensic instrument, not paperwork. Used well, it surfaces accountability problems that look like stakeholder problems, and it surfaces them in time to fix them."
Most PMs make two mistakes when they build a RACI. The first, and the more damaging, is leaving the Accountable ambiguous. Two A's, no A, "the leadership team," or "we'll figure it out together." Each is a future incident, waiting for the moment a decision has to be made and there is no agreement on who makes it. The second mistake is letting the Consulted list balloon. Every senior person who might want a say gets added on the theory that being inclusive is safer than being exclusive. The result is that no decision gets made, because by the time you have circulated a draft to twenty-three people, two have given conflicting feedback, three are on vacation, and one has changed roles.
For each decision in your RACI, you should be able to answer both in one sentence:
Most RACIs in the wild fail one of these two tests. Building one that passes both is the whole point of the template.
Most RACI spreadsheets are typed by hand, which is why so many of them have two Accountables and nobody notices for four months. A RACI matrix Excel template should not rely on the person filling it in to catch that. The sheet should catch it. Here is how this one is built.
Orientation. Rows are decisions, not tasks. Column A is the decision ID, column B the decision itself, and columns C through H are named humans in the header row. Not "Engineering." Not "the platform team." Names. Then click C4 and use View › Freeze Panes › Freeze Panes so the decision text and the name row both stay locked while you scroll a grid that gets wide fast.
Lock the cells to four letters. Select C4:H40, then Data › Data Validation › Allow: List, and type the source straight into the box:
R,A,C,I
Set Error Alert to Stop, not Warning. Warning lets people click through, and the moment someone types "A/R" into a cell every count formula below stops working. Four letters or empty. Nothing else gets in.
The formula that catches the classic error. Put this in a checker column, say column I, and fill it down the whole matrix:
=IF(COUNTIF($C4:$H4,"A")<>1,"Fix: exactly one A per row","")
That one line does more for a program than the rest of the template combined. It fires on the two-A row, which is the incident waiting to happen, and it fires on the zero-A row, which is the same incident with better manners. Add two companions in the next columns: =IF(COUNTIF($C4:$H4,"R")=0,"No R: nobody is doing the work","") and =IF(COUNTIF($C4:$H4,"C")>4,"C list too long","").
Named conditional-formatting rules. Two rules, both written from the top-left cell of the selected range so the relative references travel correctly. Select C4:H40 and add rule one, =C4="A", with a bold dark fill. The Accountable becomes the thing your eye lands on first, which is the correct visual hierarchy for a document about who decides. Then select B4:H40 and add rule two:
=COUNTIF($C4:$H4,"A")<>1
Red fill on the whole row. Note the difference between the two rules: the first uses C4 with no dollar signs so it evaluates cell by cell, the second uses $C4:$H4 so every cell in the row asks the same question and the row colors as one unit. Getting those dollar signs wrong is the most common reason a formatting rule looks broken.
The workload row nobody builds. Under the last decision, put a count of Accountables per person and drag it across the name columns:
=COUNTIF(C$4:C$40,"A")
The dollar signs sit on the row numbers so the range holds while the column moves. Now look at the totals. If one person is Accountable for fourteen of your twenty decisions, you do not have a RACI. You have a queue with that person's name on it, and the program moves at the speed of their calendar.
Printing it so people read it. Page Layout › Print Titles, set Rows to repeat at top to the header row and Columns to repeat at left to A:B. Then Scale to Fit, Width: 1 page. Without that, page two of a printed RACI is a grid of letters with no decisions and no names attached.
Excel and Google Sheets. The file is an .xlsx and works in both. COUNTIF, IF, and the formula-based formatting rules import cleanly, and the R/A/C/I dropdown becomes a Sheets data validation rule with the same four options. The one change: Excel's Stop-style Error Alert arrives in Sheets as "Reject input," which you may need to re-select, because Sheets defaults new rules to "Show warning." Re-check it after import, or the guardrail is decorative.
Five decisions from a checkout program, six weeks in. Blank means the person is Informed by default; the template keeps an explicit I only where being told matters on a schedule.
| Decision | Rivera | Okonjo | Halvorsen | Beaumont | Ilunga | Checker |
|---|---|---|---|---|---|---|
| Final launch date for guest checkout | C | R | I | A | I | |
| Payment vendor selection | R | A | I | C | C | |
| Data retention window for guest orders | R | C | I | I | A | |
| Which SKUs ship in wave one | I | C | A | I | ||
| Rollback threshold on launch night | A | R | I | A | C | Fix: exactly one A per row |
Four rows pass. One does not. Rivera and Beaumont are both marked Accountable for the rollback threshold, so the checker column fires and the conditional format reds the row. Nobody typed that error on purpose. It happens because a platform lead and a VP of Product each reasonably believe the call is theirs, and in a hand-typed grid that belief goes unexamined until launch night, at 2 a.m., with traffic dropping.
The value of the sheet is not the letters. The value is that the sheet argues back.
Decision-level. Task-level RACIs become wallpaper: hundreds of rows, nobody reads, accountability gets diluted across a sea of boxes. Decision-level RACIs are small, deliberate, and used. The template is built for decision-level.
Roughly one row per decision category that crosses team boundaries. For a typical enterprise program, that's 10 to 30 rows. If you're at 50+, you're probably writing a task list.
RACI is about decisions: who decides, who weighs in. The communication plan is about updates: who gets told what, how often. They're complements. The cheat sheet covers when each one applies.
Both. The template is .xlsx and opens cleanly in Google Sheets.
The lite version is one sheet with R/A/C/I dropdowns and the rule that matters: one Accountable per row. Free, no card.