The Decision Authority Matrix

01The Problem

Your operations manager is good. You hired them to run things. And yet a steady stream of questions still lands on you. Can I book this? Should I change that process? Is it OK to tell the client we’re moving the date?

Most of those questions have obvious answers. They know the answer. You know they know. They ask anyway, because nobody ever told them where their authority stops. Asking feels safer than guessing wrong.

The cost runs both ways. You get pulled into dozens of small calls a week, so you either micromanage or you start waving things through without looking. Then, every so often, they make a call you’d never have made, and you’re surprised. Both of you lose. They feel second-guessed. You feel like the bottleneck you hired them to remove.

This isn’t a trust problem or a capability problem. It’s a definition problem. The boundaries exist only in your head, and they shift with your mood. The fix is to write them down. Not by feel. Explicitly.

02Who This Is For

Founders handing real decisions to an operations manager, a second in charge, or a new leader, where the handover keeps stalling. If your senior people still ask permission for things they could decide, or you keep getting blindsided by calls you didn’t know were being made, start here.

It works just as well between co-founders who keep re-arguing who has final say.

03What You’ll Need

• Time commitment: 90 minutes for the first build, done together. Then a 30-minute review after 30 days.
• Prerequisites: Both people in the room, and a willingness to use real recent examples rather than hypotheticals. The matrix is only as good as the cases you test it against.
• Tools: A shared document or a whiteboard with a grid: decision areas down the side, the four levels across the top.

04The Protocol

Every decision falls into one of four levels. The definitions are short on purpose. If you need a paragraph to explain a level, nobody will use it under pressure.

• Act: Do it. No need to tell anyone. Booking a $50 Uber. Choosing which tool to use for a task.
• Act & Inform: Do it, then tell the founder. Changing the order of the week’s priorities. Adjusting a process that affects the team.
• Propose: Bring a recommendation. The founder decides. Taking on a new supplier. Setting a new company-wide policy.
• Escalate: Don’t touch it without the founder. Signing a major contract. Changing the strategic direction.

The instinct is to put everything in Escalate. Resist it. Escalate is a queue, and you are the only person serving it. The whole point of the matrix is to move as much as possible to the left.

Decisions come up in the same handful of areas again and again. List them down the side of the grid. Start with these seven and add your own:

• Budget: what can they spend, and at what amount does it change level?
• Scope: what can they add, cut or change in a piece of work?
• Timeline: which dates can they move, and which are fixed?
• People: rosters, workloads, performance conversations, hiring.
• Stakeholder communication: what can they tell clients, suppliers and the team without you?
• Process: which ways of working can they change?
• Risk: what exposure can they accept on the company’s behalf?

For each area, fill in all four columns. Be specific. Use dollar amounts, real scenarios and recent examples. “Small spends” is not a boundary. “Under $500, Act. $500 to $5,000, Act & Inform. Over $5,000, Propose” is.

Go through each area with the founder and ask one question:

“What would need to happen before you’d feel blindsided?”

That threshold is the line where Act & Inform becomes Propose. Below it, the founder is happy to hear about it afterwards. Above it, they want a say before it happens.

Founders usually set this line too low the first time. Push on it. “Would you really want to be asked before that?” often moves the line by an order of magnitude.

Now go through the same areas with the person taking on the authority, and ask:

“Where do you currently ask permission when you probably already know the answer?”

Those are the decisions to push down to Act or Act & Inform. You’ll be surprised how long this list is. Most of the questions landing on the founder’s desk are already answered in the operator’s head. They’re asking for cover, not direction.

Walk through the last few weeks together. Where did the operator hesitate and wait? Where was the founder surprised by something that had already happened? Put each real example onto the matrix and check that it lands where you both expect.

When the two of you disagree on where something belongs, that’s the most useful moment in the session. Talk it through until you agree, then write the answer in the grid. Every one of those disagreements was going to surface later anyway, mid-week and under pressure.

For decisions that don’t fit neatly in a box, use one filter: if this turns out to be wrong, what does it cost to fix? Cheap to undo means Act or Act & Inform. Expensive or impossible to undo means Propose or Escalate. That’s the one-way door and two-way door test from the Velocity Protocol, applied to who decides.

It will go wrong. That’s part of the deal.

When the operator makes a call that misses, they own it. They tell the founder what happened and what they’d do differently. The founder’s question is not “Why did you do that?” It’s “What did I not define clearly enough?” A wrong call that’s cheap to undo is cheap to fix. An operations manager who’s frozen is expensive every single day.

Revisit the matrix after 30 days. Some Propose items will be ready to move to Act & Inform. A few boundaries will need tightening. That’s the system working, not failing. Both of you sign off on the new version. Both of you own it.

05What You’ll Find

The first thing that changes is the volume of questions. Once the operator can see in writing that a decision is theirs, most of the “quick question” messages simply stop. Founders usually get several hours a week back in the first month.

The second change is in the operator. People don’t grow into authority they’ve only been half given. When the boundary is written down, they stop checking and start deciding, and their judgement improves quickly because they’re finally getting to use it.

The third change is quieter. You’ll find out where your own boundaries were unclear, or kept changing. The matrix makes you decide once what you’ve been deciding over and over, case by case. That’s often the most useful thing in the whole exercise.

The pattern I see most often: within a few weeks, the operator is making most of their area’s calls without checking in, and the founder gets a short weekly summary instead of a running stream of questions. The matrix doesn’t make anyone more capable. It makes clear which decisions are already theirs.

06Adaptations

Co-founders: Use the same grid, but fill it in independently first, then compare. The gaps between your two versions are as revealing as the overlaps. If one of you puts hiring under Act and the other puts it under Escalate, you’ve found your next argument before it happens.

A whole leadership team: Build one matrix per leader, not one for the team. Authority sits with a person, not a function. Share them with the team so everyone knows whose call is whose.

A new hire: Start tighter, with more in Propose, and say up front that the plan is to loosen it at 30 and 60 days. That way, widening their authority reads as progress rather than a correction.

Solo founder with contractors: Keep it to one page. Budget, scope and client communication are usually the only areas that matter.

07Where This Came From

I built this for founders handing real decisions to an operations lead for the first time. The same pattern kept showing up: a capable operator asking permission, and a founder still pulled into the day-to-day. The four levels are a simplified version of the delegation levels used in many management frameworks, cut down to the four that hold up when you’re busy.

The “what does it cost to fix” filter comes from Jeff Bezos’s one-way and two-way door distinction in his 2015 Amazon shareholder letter. The questions for each person came from noticing that founders and operators draw the line in different places for different reasons. The founder draws it at surprise; the operator draws it at safety. The matrix only works when both lines end up in the same place.