Deciding vs executing

There are two activities that owners conflate more than any others, and the conflation is the single biggest blocker to delegating work, hiring well, and using AI effectively.

The two activities are deciding and executing. They look similar from the outside. They feel similar to the person doing them, especially when one person does both at once. They are, underneath the surface, completely different kinds of work, and a business that does not separate them ends up with the owner doing both forever — because the owner is the only person who can do the deciding part, and the deciding got tangled up in the executing.

This piece is about the difference. What deciding actually is. What executing actually is. Why owners conflate them. And what becomes possible the moment they get pulled apart.

The two activities

Executing is doing the work. The labor. The action. The thing that happens. Drawing the patio plan. Writing the email. Sending the invoice. Fixing the mower. Calling the customer back. Delivering the materials. Pouring the foundation.

Deciding is choosing what gets done, by whom, when, with what trade-offs. Whether to take this job. What price to charge. Whether to hire this person. Whether to chase this lead or let it go. Whether to invest in this piece of equipment. How to handle this customer's complaint. Whether the work is good enough to send.

In a small business, both activities happen all day, often in the same person. The owner walks into the shop, sees a pile of work, decides what gets done first (deciding), then does some of the work themselves (executing), then a customer calls and the owner decides whether to take the call now or later (deciding), then the owner takes the call and walks the customer through their options (executing the call, while also deciding what to recommend), then the owner reviews a quote their estimator wrote (deciding whether it is good enough), then signs it (executing the action of signing), then sends it (executing the send).

Inside that day, the deciding and the executing are interleaved so closely that the owner experiences it as one continuous stream of work. It is not. It is two different kinds of work, with completely different requirements, that happen to be living in the same body.

Why this matters

The reason it matters is that deciding and executing have completely different leverage profiles.

Executing scales with hands. If you have ten people, you can execute ten times more work than one person. Add another five people and you can execute fifteen times more. Executing is linearly bounded by capacity. The way to do more executing is to add more people, more time, or more tools.

Deciding does not scale that way. A decision is bounded by judgment and information, not by hands. One owner with good information can make a hundred decisions a day. Adding ten more "deciders" does not produce ten times the decisions — it produces conflict, confusion, and slow consensus. Deciding scales by making good decisions reusable, not by adding decision-makers.

This is why the highest-leverage move an owner can make is to separate the deciding work from the executing work, and figure out which decisions are truly load-bearing. The load-bearing decisions stay with the owner (or the small group of leaders who hold them). The executing work, and the routine decisions that come with it, get systematized and delegated.

When this separation happens cleanly, three things become possible:

• The executing work scales by adding people and tools, without losing the judgment that made the business work in the first place.
• The owner's time gets reclaimed for the decisions that actually require their judgment.
• AI becomes usable, because AI can take on a chunk of the executing — drafting, retrieving, classifying, summarizing — without ever crossing into the deciding work, which the owner keeps.

When the separation does not happen, the opposite. Every decision and every execution stay tangled in the owner's head. Hiring does not relieve workload because new hires cannot do anything without a decision the owner has not separated out. Software does not relieve workload because the software cannot tell which steps require judgment and which do not. AI does not relieve workload because the AI ends up either taking decisions it should not or being kept so far from the work that it does nothing useful.

Why owners conflate the two

Owners who have built a small business by themselves do not separate deciding and executing because, for the first phase of the business, they do not need to. They are doing both, all day, every day. The fact that the two are different kinds of work is academic when one body is doing both.

The conflation becomes a problem at the moment the business starts to outgrow one body. The owner tries to delegate. The delegation does not stick. They try again. The new person needs to be told what to do every step of the way, because the deciding has never been separated from the executing. Every executing task brings a deciding task with it, and every deciding task brings the owner back into the loop.

The owner concludes "I just can't find good people." Or "this software doesn't work for our business." Or "AI isn't ready yet." None of those are quite right. The actual problem is that the owner has not yet untangled the two activities, and until they do, no person and no system can take work off their plate.

What deciding looks like, separated out

To untangle, you have to be able to look at any task in your business and identify the deciding part of it.

Take a quote. Writing a quote is mostly executing — pulling the right line items from the catalog, calculating the totals, formatting the document. But there are deciding moments inside it: which line items apply to this particular job, what markup to use given the customer relationship, whether to include the optional add-ons, whether the price feels right or not. The executing is most of the time. The deciding is most of the judgment.

Take a customer complaint. Responding to a complaint is mostly executing — listening, taking notes, communicating, documenting. But there are deciding moments: whether this complaint is reasonable, whether to offer a credit, whether to escalate, whether to fire the customer.

Take a hire. The hiring process is mostly executing — posting jobs, scheduling interviews, taking notes, sending offers. The deciding is small in volume but enormous in weight: which candidate, what role, what compensation, what risk to take.

The pattern: in almost every task in a business, the executing is most of the work and the deciding is most of the value. Owners who can see this pattern in real time, on the work in front of them, can start to separate.

What executing looks like, separated out

The flip side: most of what an owner spends their day on is executing — work that does not require their specific judgment, but happens to be on their plate because nobody else has been set up to do it.

Reading every email that comes in. Most of that is executing. The deciding is which ones need a response, which can wait, and what tone to take. The reading itself is just hours of work that other systems and people could absorb.

Writing every quote. Most of that is executing. The deciding is which jobs to bid on at all, and how to position the price. The writing is mostly mechanical.

Approving every invoice. Most of that is executing — verifying the numbers add up, checking that the work was done. The deciding is whether anything looks anomalous enough to warrant a closer look.

Owners are often shocked when they audit how their day actually breaks down. They thought they were doing strategy. They were mostly doing executing — small executing tasks, repeated hundreds of times, that they had absorbed because they could not figure out how to separate the executing from the (much smaller volume of) deciding inside each one.

How AI changes this conversation

AI is the first technology that can take real chunks of executing work without taking the deciding work. That is what makes it different from previous waves of business software, and that is also what makes it dangerous when an owner has not separated the two clearly in their own mind.

AI can draft the quote. The owner decides whether to send it.

AI can summarize the customer email thread. The owner decides what to do about it.

AI can classify the incoming inquiries by likely fit. The owner decides which ones to pursue.

AI can extract the line items from the contract. The owner decides whether to sign.

The pattern is the same in every case. AI takes the executing. The owner keeps the deciding. The handoff between them is the line we keep referring to in other pieces — the wall between AI drafts and a person decides — and it is built into the technology's strengths and weaknesses for reasons that are not going to change soon.

The owner who has separated deciding from executing in their own mind can use AI to take significant chunks of executing off their plate — and the time freed up gets reinvested into deciding the things that actually require their judgment. The owner who has not separated the two will reach for AI to take "the whole task" off their plate, and will end up either being disappointed (the AI cannot decide) or in trouble (the AI did decide and the decision was wrong).

How to start separating

You do not need a workshop or a consultant to start. You need an honest pass through your own week.

Step one: at the end of a workday, list the things you did. Not every keystroke — the actual chunks of work. Wrote three quotes. Replied to a customer complaint. Sat in a planning meeting. Approved invoices. Reviewed the schedule. Took two phone calls.

Step two: for each item, ask: which part of this was deciding, and which part was executing? A quote — most of it executing, with two or three small decisions inside. A customer complaint — mostly executing the response, with one or two decisions about how to handle. The planning meeting — mostly deciding. The invoices — mostly executing the verification, with one or two judgment calls.

Step three: total up the deciding time across your day. It will be much smaller than you expected. Most owners' days, honestly examined, are 80% executing and 20% deciding — sometimes more lopsided. The 20% is what only you can do. The 80% is what you have absorbed, and what could in principle move.

Step four: pick one chunk of executing, the one that is taking the most of your time, and figure out who or what could absorb it. A person? A system? An AI? Some combination? The first separation is the hardest. After that, it gets easier, because you know what you are looking for.

This work is uncomfortable for most owners, because the deciding/executing distinction makes visible something most owners have spent years not looking at: how much of their day is mechanical work that they have been carrying alone. Once you can see it, you can move it. Until you can see it, you cannot.

The owner's job, properly defined

The owner's actual job, in a healthy business, is to make the decisions only they can make — and to set up the people, systems, and tools that absorb everything else.

That is not "do less work." It is doing the right work. The decisions about the business — what direction it goes, what risks it takes, what kind of customers it serves, what price it charges, what it stands for — are the decisions an owner is uniquely positioned to make. The executing of the day-to-day, the routine decisions inside the executing, the work of running the loop one more time — that is what the rest of the business should be set up to handle.

When the separation is clean, the owner has time to think. To watch. To notice. To make the bigger decisions well, because the smaller ones are not constantly demanding attention. The business runs more reliably, because the routine decisions are happening through systems instead of through a single overloaded person.

When the separation is not clean, the owner is buried. The business runs at the speed of the owner's bandwidth. The owner cannot take a vacation, cannot take on bigger work, cannot grow, because every part of the operation requires their judgment for things that should not require their judgment.

The work of pulling these two apart is the work that turns a busy founder into the operator of a real business. It is the most consequential mental shift we know, and it is the precondition for almost every other improvement — including, but not limited to, AI.

For more on the building blocks, read What a business actually is, Roles, not people, Source of truth (and what it costs to lack one), and Workflows: how value moves through a business. For where this fits with AI, read What AI is for, and what it isn't and An AI system, walked through like a building.