Overview

A RACI matrix answers a deceptively simple question: who does what? It sounds easy. It's not. In most organizations, the answer is mushy. "Finance reviews the budget." Okay, but Finance has four people. Does one of them review it or all of them? Can any of them approve it, or only one? What if that person is out? There's always ambiguity, and that ambiguity costs time and creates conflict.

A RACI matrix removes the ambiguity. For every decision or task, you specify: who's Responsible (does the work), who's Accountable (owns the outcome), who should be Consulted (has input), and who should be Informed (needs to know it happened).

The problem: Creating a RACI matrix requires you to interview everyone, document the current state, think through the ideal state, and make trade-off decisions about who should own what. That's days of work. AI can help you draft it in hours. But AI makes predictable mistakes in RACI matrices, and you need to know what to watch for.

RACI Basics (Quick Refresher)

R = Responsible: The person or role who does the actual work. There's usually one R, sometimes multiple (if work is shared).

A = Accountable: The person or role who owns the final decision or outcome. There can only be one A. If something goes wrong, this person is responsible for fixing it or explaining why. A is often a manager or owner level.

C = Consulted: People whose input matters. They're asked for their perspective before the decision is made. Not everyone needs to be consulted on everything. Pick the ones who have specific expertise or perspective. This often includes subject matter experts.

I = Informed: People who need to know the decision or outcome happened. They don't have input, but they need to be in the loop. This can be a broad group (all staff), or a narrow group (specific departments).

Example: "We need to decide on new project management software."

• R = Project Manager (will do the evaluation and create shortlist)
  - A = VP of Operations (makes final call, owns the decision)
  - C = Finance (cost impact), IT (integration concerns), project team leads (will use the tool)
  - I = All staff (everyone needs to know what was chosen and when it starts)

Why this matters: You're not asking Finance to decide. You're asking Finance for input. You're not asking project leads to decide. You're telling them the decision was made. Clear boundaries.

The RACI Prompt: Getting AI to Do the First Draft

To get AI to create a reasonable RACI matrix, you need to provide:

• List of activities/decisions that need coverage
  - List of roles/functions involved
  - Context about your organization (centralized vs. distributed, how decisions are made, any constraints)
  - Specific guidance where you know the right answer

Template:

Role: You are an organizational design specialist who creates RACI matrices.

Context:
- Organization: [Size, structure, decision-making style]
- Focus: [What decisions/activities does the matrix cover]
- Roles involved: [List all roles that might be involved]
- Decision-making style: [Centralized? Consensus? Delegated?]
- Constraints: [Any non-negotiables about who owns what]

Task: Create a RACI matrix that clarifies who's responsible for each activity.

Format: Table with:
- Rows: List of activities/decisions
- Columns: Each role
- Cells: R (Responsible), A (Accountable), C (Consulted), I (Informed), or blank

Guidelines:
- Every activity should have exactly one A
- Only mark R/C/I where they add value; blank cells mean not involved
- Avoid marking too many people as C (consulted); reserve that for people whose input actually matters
- Think about escalation: if someone's assigned R, can they escalate to their manager (A)?

Assumptions to verify:
- [List 2-3 specific assignments you're unsure about and want the AI to think through]

Real example prompt:

Role: You are an organizational design specialist who creates RACI matrices for operations teams.

Context:
- Organization: 60-person manufacturing company. We have VP of Ops, 3 Operations Managers, Finance Manager, HR Manager, 4 Department Leads (Sales, Production, Quality, Logistics).
- Focus: Day-to-day operations decisions (hiring, vendor management, capacity planning, process changes, emergency decisions)
- Roles involved: VP Ops, Operations Manager, Finance Manager, HR Manager, Department Leads
- Decision-making style: Generally delegated—managers make decisions within their area. VP escalates exceptions.
- Constraints: HR decisions should stay with HR Manager. Finance decisions should have Finance Manager involved.

Task: Create a RACI matrix for our key operations activities. Show who's responsible, who decides, and who should be consulted.

Format: Table with activities as rows, roles as columns. Use R/A/C/I or blank.

Activities to include:
1. Hiring new operations staff
2. Vendor selection and contracting
3. Process improvements (department-level)
4. Emergency capacity/resource decisions
5. Capital equipment purchases
6. Headcount planning and budgets
7. Performance management (ops staff)
8. Cross-department coordination meetings

Assumptions to verify:
- Should all vendor contracts be approved by Finance Manager, or only over $X threshold?
- Who approves process improvements—the Department Lead or VP Ops?
- For emergency capacity decisions, can an Operations Manager decide alone or do they need VP approval?

What AI generates (first pass):

OPERATIONS RACI MATRIX

Activity | VP Ops | Ops Manager | Finance Mgr | HR Manager | Department Lead
─────────────────────────────────────────────────────────────────────────────
Hiring new ops staff | A | R | C | A | C
Vendor selection/contract | A | R | A | | C
Process improvements (dept) | C | R | | | A
Emergency capacity decision | A | R | | | C
Capital equipment purchases | A | | A | | R
Headcount planning/budgets | A | R | A | C | C
Performance management (ops) | C | R | | A |
Cross-dept coordination | A | R | | | I

Your review: Some issues jump out immediately:

• Row 1 (Hiring ops staff) has TWO A's (VP Ops and HR Manager). That violates the RACI rule—there can only be one A.
  - Row 2 (Vendor selection) also has TWO A's. Which one actually owns the decision?
  - Row 5 (Capital equipment) shows Department Lead as R, but capital equipment is usually a VP or Finance decision. This seems backwards.
  - Row 7 (Performance management) shows HR Manager as A, but Operations Manager is the direct manager. Should ops manager own this?
  - Row 6 (Headcount planning/budgets) has both Finance Manager and VP Ops as A. Who decides if we can hire?

Your iteration request:

I see some issues. Let me clarify:

1. Hiring ops staff: HR Manager is Accountable (they make hiring decisions). Operations Manager is Responsible (sources candidates, runs interviews). VP Ops is Consulted (approves fit with ops needs). Revise accordingly.
2. Vendor selection: Finance Manager is Accountable (owns the final contract decision). Operations Manager is Responsible (evaluates vendors, negotiates). VP Ops is Consulted (large contracts). Department Leads are Informed.
3. Capital equipment purchases: This should be VP Ops Accountable (final decision on spend), Finance Manager Consulted (budget impact), Department Lead Responsible (identifies need, evaluates options).
4. Performance management of ops staff: Operations Manager is Accountable and Responsible (direct manager relationship). HR Manager is Consulted (policy, escalations). VP Ops is Informed.
5. Headcount planning/budgets: VP Ops is Accountable (decides headcount strategy), Finance Manager is Consulted (budget constraints), Operations Manager is Responsible (identifies staffing needs).
6. Add one more row: "Vendor contract changes mid-contract" — this is different from initial selection. Add a row: "Vendor contract changes (mid-contract)".
7. Add another row: "Compliance audits and corrective actions" — important for ops.

Updated RACI matrix (revised):

OPERATIONS RACI MATRIX (REVISED)

Activity | VP Ops | Ops Manager | Finance Mgr | HR Manager | Department Lead
─────────────────────────────────────────────────────────────────────────────
Hiring new ops staff | I | R | | A |
Vendor selection/contract | C | R | A | | C
Vendor contract changes | | R | A | | C
Process improvements (dept) | C | C | | | A
Emergency capacity decision | A | R | | | C
Capital equipment purchases | A | R | C | | R
Headcount planning/budgets | A | C | C | I | C
Performance management (ops) | I | A/R | | C |
Cross-dept coordination | A | R | | | I
Compliance audits/corrective | C | R | C | C | A

Much better. Now the accountabilities are clear. For hiring, HR Manager decides. For contracts, Finance Manager decides. For capacity, VP Ops decides. For compliance, Department Lead is responsible with VP Ops consulted. Clear escalation paths and clear who owns what.

The two biggest AI mistakes in RACI matrices

Mistake 1: Creating multiple A's (Accountable). There must be only one Accountable per activity. AI doesn't enforce this and will generate matrices with two or three A's in a row. You have to fix it. This is a structural problem that creates confusion. When there are two A's, neither person feels accountable, and decisions don't get made clearly.

Mistake 2: Marking C (Consulted) too broadly. AI thinks everyone should have a voice. But C should be reserved for people whose specific expertise matters. If you mark 5 people as C for one activity, none of them feel their input is valued, and the decision gets slowed down. Be selective. The rule: would this person say "wait, you need my input on this," or are they just interested?

Common AI Mistakes in RACI Matrices and How to Catch Them

Mistake #1: Multiple Accountables

This happens most often when two roles share a function. Example: "Hiring should be shared between HR and Operations Manager." So AI marks both as A. But that's not a RACI matrix—that's a conflict waiting to happen. You need to pick one.

Real-world impact: A new hire doesn't work out. HR Manager says "Operations Manager should have been more selective." Operations Manager says "HR gave me the final candidate—that's on HR." Nobody owns the problem. Nobody fixes it.

How to catch it: Read through the A column. Any activity with more than one A is wrong. Ask yourself: if something goes wrong with this activity, who gets called into the room to explain? That's your A.

Mistake #2: Over-consulting

AI marks everyone as C. It thinks more input is better. So you get: "Vendor selection decision. Consulted: Finance, Operations, Product, Sales, HR, Quality, Safety, randomly someone from accounting."

Real-world impact: Vendor selection takes three weeks of back-and-forth meetings instead of one week, because you're trying to get input from 8 people instead of 3. Each person thinks their input matters. When the decision goes a different direction, they feel unheard.

How to catch it: If you have more than 2-3 people marked as C for any activity, you're consulting too much. Ask: "Do I actually need their input, or just need them to know what was decided?" If it's the latter, mark them I instead of C.

Mistake #3: Mismatch Between R and A

You mark someone as Responsible who reports to someone other than the Accountable. Example: Marketing Manager is A, but an external contractor who doesn't report to Marketing Manager is R. That's broken. R should either report to A, or there's a clear escalation path.

Real-world impact: The contractor makes a decision the Marketing Manager doesn't agree with. Marketing Manager can't override it because they don't have authority. The contractor doesn't feel accountable to Marketing Manager because they don't report to them.

How to catch it: For each activity, check: does R report to A? If not, is there a clear escalation path? If neither, there's a structural problem that needs fixing.

Mistake #4: Consultant as Accountable

AI assigns Consulted and Accountable to the same role. "Legal Consulted and Accountable for contract approval." That doesn't make sense. If someone's accountable, they should own the decision. If they're consulted, they provide input but someone else owns it. These are contradictory.

Real-world impact: Legal feels like they have veto power on contracts, but the Procurement Manager feels they own the decision. Conflict happens. You can't tell who actually decides.

How to catch it: Make sure one role is A and different roles are C. If the same person is both, clarify which is it—do they own the decision, or do they just advise?

Mistake #5: Unclear Threshold for Escalation

AI creates a RACI matrix without noting when decisions escalate to the next level. Example: Manager is A for hiring. But is that true for all hires, or only hires under $60K salary? If it's conditional, that needs to be noted. The matrix looks like there's one rule, but actually there are multiple rules depending on the situation.

How to catch it: For each A assignment, ask: "Is this always true, or only under certain conditions?" Add footnotes where conditions apply.

The RACI Verification Checklist

After AI generates a RACI matrix, before you publish it, run through this checklist:

Rule Compliance:

• Every activity has exactly one A (Accountable) — not zero, not two?
  - Every activity that has an R (Responsible) also has an A?
  - No one is both C and A for the same activity (unless explicitly intended and documented)?
  - No activity is all blank (someone should be involved)?
  - For cross-functional activities, are all relevant functions involved?

Clarity:

• Could someone new to the org read this and understand who owns what?
  - Are there any columns so empty that the role seems to have no responsibilities? (Might mean they're not needed, or might mean underutilization.)
  - Are there any rows with so many R's that it's unclear who's really responsible? (Might mean the activity is shared, which is fine—but it needs to be explicit.)
  - Would each person on the team understand what their role is in each activity?

Operationality:

• For each A, do they have the authority to make the decision, or do they need to escalate?
  - For each C, is their input actually needed, or are they marked out of an abundance of caution?
  - Are there any activities that are marked I (Informed) where someone should really be C?
  - Are there decision thresholds where escalation happens? (e.g., "Manager approves hires under $X, VP approves over $X")

Completeness:

• Are there critical activities missing from the matrix?
  - For cross-functional activities, are all relevant functions represented?
  - Are escalation rules clear (e.g., if Ops Manager encounters X situation, they escalate to VP Ops)?
  - Have you thought about who would be involved in crisis/escalation scenarios?

Try This Now: Drafting and Fixing a RACI Matrix for Different Scenarios

Scenario 1: Product Development Team (SaaS)

You're building a new product for your SaaS company. You need a RACI matrix for all the decisions involved in product development (roadmap, feature prioritization, technical design, launch timing, pricing).

Roles involved:

• VP Product (owns product strategy)
  - Product Manager (manages this specific product)
  - Head of Engineering (owns technical feasibility)
  - CFO (owns pricing, revenue impact)
  - CMO (owns launch marketing)
  - CEO (ultimate owner, escalation point)

Your prompt:

Role: You are an organizational design specialist.

Context:
- Organization: SaaS company with 50 people. Flat structure, but with clear functional leads.
- Product: We're launching a new product line. I need a RACI matrix for all decisions in development cycle.
- Roles: VP Product, Product Manager, Head of Engineering, CFO, CMO, CEO
- Decision-making: VP Product and Head of Engineering are co-equal. CEO has final say on anything. Marketing has strong input on launch strategy.

Task: Create a RACI matrix for new product development decisions.

Activities:
1. Product roadmap decisions (what gets built)
2. Feature prioritization
3. Technical design and architecture decisions
4. Pricing strategy
5. Launch timing decision
6. Go/no-go decision before launch
7. Launch marketing strategy
8. Post-launch adjustments
9. Customer feedback prioritization
10. Scope/feature cuts if we're behind schedule

Format: RACI table with one A per activity. No blanks unless truly not involved.

Key constraints:
- VP Product should be A on roadmap and features (product domain)
- Head of Engineering should be A on technical decisions (tech domain)
- CFO should be A on pricing
- CMO should be Consulted (not Accountable) on marketing—PM owns the launch
- CEO is informed but not directly involved unless escalated

AI generates (first pass):

PRODUCT DEVELOPMENT RACI

Activity | VP Product | PM | Head Eng | CFO | CMO | CEO
─────────────────────────────────────────────────────────
Roadmap decisions | A | R | C | | | I
Feature prioritization | A | R | C | | C |
Tech design | | R | A | | |
Pricing | C | | | A | | I
Launch timing | | A | C | C | C | I
Go/no-go launch | | A | C | A | C |
Launch marketing | | | | | A | I
Post-launch adjustments | A | R | C | | C | I
Customer feedback prioritization | | A | C | C | I |
Scope/feature cuts | A | | C | | | I

Your review: Issues found:

• Row 2 (Feature prioritization): VP Product is A, but shouldn't Head of Engineering be Consulted (they'll know feasibility constraints)?
  - Row 4 (Pricing): CFO is A (correct), but shouldn't VP Product be Consulted (owns the product strategy and value prop that drives pricing)?
  - Row 5 (Launch timing): PM is A. Is that right? Shouldn't VP Product be involved given it affects strategy?
  - Row 6 (Go/no-go launch): Two A's (PM and CFO). Only one should be A. Who ultimately decides? Probably VP Product or CEO.
  - Row 7 (Launch marketing): CMO is A. But CEO and VP Product should be Informed, not blank.
  - Row 8 (Post-launch adjustments): Is CEO really Informed? Or should they only hear about major issues?

Your refinement:

Fix these:

1. Feature prioritization: Mark Head of Engineering as C (they need to weigh in on feasibility).
2. Pricing: Mark VP Product as C (their product strategy informs pricing).
3. Launch timing: Mark VP Product as A (this is a strategic call, not just a project management task).
4. Go/no-go launch: This should be CEO A (final decision on launch). PM is R (executes launch). VP Product and Head of Engineering are C (they advise on readiness). CMO is I.
5. Launch marketing: Mark CEO and VP Product as I (they need to know when launch happens).
6. Post-launch adjustments: Mark CEO as blank or I (only informed of major changes, not all adjustments).
7. Customer feedback prioritization: This is VP Product's job, not PM. Mark VP Product as A or make it PM A with VP Product as C.

Additional considerations:
- When does CEO override? Add a note: "CEO has final say on any decision where A disagrees with Head of Engineering"
- Add escalation rule: "Feature cuts that affect more than 20% of roadmap go to CEO"

Updated matrix (after fixes):

PRODUCT DEVELOPMENT RACI (REVISED)

Activity | VP Product | PM | Head Eng | CFO | CMO | CEO
─────────────────────────────────────────────────────────
Roadmap decisions | A | R | C | | | I
Feature prioritization | A | R | C | | |
Tech design | | | A | | |
Pricing | C | | C | A | |
Launch timing | A | R | C | | |
Go/no-go launch | C | R | C | | | A
Launch marketing | | C | | | A | I
Post-launch adjustments | A | R | C | | |
Customer feedback prioritization | A | R | C | | |
Scope/feature cuts | A | I | C | | | C (escalation if >20%)

Notes:
- CEO has final say on any decision where Accountable and Head of Engineering disagree
- CFO is consulted on pricing, not the main A
- All product decisions flow through VP Product; PM executes

Now every activity has one clear owner (A), and the matrix actually reflects your decision-making structure.

Scenario 2: Operations Team (Manufacturing)

You manage operations for a 100-person manufacturing company. You have Operations Director (you), Plant Manager, Maintenance Manager, Quality Manager, Finance Manager, HR Manager, Sales Manager.

Key activities: Capacity decisions, equipment maintenance scheduling, quality improvements, vendor management, staffing decisions, safety incidents, process changes, budget decisions, new equipment purchases.

Build a RACI for these using the same process above. The key difference from the SaaS example: manufacturing has more compliance requirements, more safety considerations, and more vendor dependencies.

Pro tip: After building your RACI matrix, do a test run. For the next three decisions your team makes, use the RACI to guide them. "According to the matrix, Finance Manager is A for this. Let's follow the process." You'll quickly learn whether the matrix matches reality. If it doesn't, adjust. After one month of real-world use, you'll have a matrix that actually reflects how you work.

What to Do Monday Morning

• List the top 10 decisions that get made in your area. What decisions create confusion or conflict today? Start with those.
  - List all roles that should be involved in those decisions. Include roles you report to and roles that report to you. Don't forget cross-functional roles (Finance, HR, Compliance).
  - Create a prompt asking AI to draft a RACI matrix. Use the template in this lesson. Be specific about your organization and decision-making style.
  - Review the AI output for the five common mistakes. Specifically look for multiple A's, over-consulting, mismatches between R and A, and unclear escalation thresholds.
  - Iterate with AI to fix problems. Tell AI what to change: "Remove the second A," "Add this person as Consulted," "Change escalation threshold."
  - Walk the matrix with your team. Ask: "Is this how we actually make decisions, or how we should make decisions?" If there's a big difference, note it.
  - Publish the verified matrix. Share it with your team. Tell them: "This is who owns what. When you have a question, come to the A. This is our decision framework."
  - Live with it for a month. When decisions come up, use the matrix. Track whether it works. After a month, refine based on experience.

Key Takeaways

• Every activity must have exactly one Accountable. This is the RACI rule AI doesn't always follow. Watch for it. It's your most critical check.
  - Over-consulting is a common mistake. Mark people as C only if their specific input affects the decision. Otherwise, mark them I. Each C should feel that their input matters.
  - Responsible should report to Accountable. If not, there's a structural problem that should be addressed. You can't hold someone accountable for work done by someone who doesn't report to them.
  - Escalation thresholds should be explicit. "Manager is A for hiring" is incomplete if some hires escalate to VP. Note the threshold.
  - Use RACI to clarify, not to constrain. The goal is to remove ambiguity and conflict. If the matrix feels overly rigid, adjust it.
  - A RACI matrix changes when roles change. When you hire someone new or restructure, update the matrix. Don't let it get stale.
  - Walk the matrix with your team before publishing. Their input is valuable. A RACI that people don't agree with won't work.
  - RACI is a team conversation tool, not an org chart. It clarifies decisions, not just roles. Use it to talk about how your team actually works.

FAQs

What if the Accountable and Responsible are the same person?

That's fine. Sometimes one person owns and executes an activity. Just make sure they have the authority and time to do both. Mark both A and R for that role. This is common for small teams or specialized functions.

Can someone be both Responsible and Consulted?

Hmm, that's a bit redundant. If someone's R (doing the work), they already have their input built in. Marking them C too is overstating it. Keep R and let the other roles be C. The exception: if someone does the work AND provides expert input on someone else's work, they might be R on one activity and C on another activity.

What if a decision is truly shared between two people?

Then mark one as A (tie-breaker if they disagree) and one as R (does the work). Or, if they truly co-own it, pick the more senior one as A and note that both need to agree. Add a footnote if needed. But RACI expects one A, so be explicit about the co-ownership arrangement.

How often should I update the RACI matrix?

Any time your roles or responsibilities change significantly. If a new function is added, or a role's scope changes, update the relevant rows. Keep it current. A stale RACI is worse than no RACI. Review it quarterly at minimum.

Should I use RACI for small teams?

If you have 4-5 people, maybe RACI is overkill. You probably already know who owns what. But if you have 10+ people or cross-functional dependencies, RACI removes ambiguity. It's worth doing. Even small teams benefit from writing down the decision framework once, to prevent future confusion.

What about shared services or support functions?

Support functions (IT, Finance, HR) are usually C or I on decisions in other departments, not R or A. They provide input or expertise. Be clear about this. If IT is marked as R on something, they actually do the work. If they're C, they advise. Don't mark them R if they just help.