What 15 years of founder/CTO work taught me about operator-mode

I have shifted between founder mode and operator mode more than once over the last fifteen years. Early in the arc I carried the same habits across the line. The habits worked in founder mode. They burned the operator chair down. I had to learn that the work uses opposite muscles, not adjacent ones, and the reflex that wins one is the reflex that loses the other.

Founder mode is deciding what gets built. Operator mode is deciding what doesn’t.

#The split

Founder mode is generative. You stand at the front of an empty room and the job is to populate it with the right things: the right product, the right team, the right customers, the right narrative. The bias is toward action, because nothing exists yet and the cost of inaction is much higher than the cost of being slightly wrong about what to build. You move fast, decide a hundred small things a day, and revise as you learn. Most days you can’t tell whether you’re winning, but you can usually tell whether you’re moving.

Operator mode is selective. The room is full. The job is to make sure the right things stay and the wrong things leave. The bias is toward saying no, because anything you say yes to gets built, and the cost of building the wrong thing now compounds across a team that has to maintain it. You move slower, make fewer decisions per day, and the decisions you make matter more. The score is unclear; the calls you got wrong show up in three months in a calendar of meetings that didn’t need to happen.

#What actually changes when you shift

Three things.

Calendar shape. In founder mode I spent most of my week building or talking to customers. Two or three hours of meetings a day, no more. In operator mode the calendar is the work. Five hours of meetings is a light day, because the meetings are where the team negotiates which things stay and which things leave. A founder hearing that thinks that sounds awful, and is correct. It is awful as a founder. It is the job as an operator.

Decision count. Founders make dozens of decisions a day, most of them small, most of them reversible. Operators make a handful a week, most of them hard to reverse without a real cost to people on the team. The number went down by an order of magnitude when I shifted. The weight went up by an order of magnitude. The first time I noticed I’d been running a quiet week was the moment I realized I’d quietly made three decisions in three meetings, and that was the week.

Information shape. In founder mode information came through whichever channel was fastest: the customer call, the bug report, the analytics dashboard. In operator mode information comes through people. The team is the filter. If the team is calibrated, the read is clear. If the team is not calibrated, the read is distorted, and the calibration is the operator’s actual job.

#Four habits I had to break

Each one worked for me in founder mode. Each one stopped working at the operator boundary.

1. Fix it yourself. As a founder, the fastest path from “something is broken” to “something is shipped” was usually me. The first time I tried this from the operator chair, I rewrote three customer-facing bugs at midnight instead of paging the engineer who owned them. The bugs got fixed. The engineer noticed. The team noticed who noticed. Now I name the owner first and let the owner work.

2. Fastest path to the answer. Founders optimize for getting the answer. Operators optimize for the team learning the answer. The slower path through the team usually produces a worse answer once and a better team forever.

3. The strong opinion in the room. As a founder, the strongest opinion in the room is usually yours, and that’s fine: the company is small enough that the opinion is the strategy. As an operator, the strongest opinion in the room is a self-silencing device. The whole meeting becomes a search for what the loudest person already decided. Now I ask the question first, listen second, and put my opinion in the room last, if at all.

4. The all-hands tactical. Founders use the all-hands to coordinate tactically: this week we’re doing X, here’s what to expect. Operators use the all-hands for the longer arc — where we are, why this quarter matters, what changes if we win it. The tactical work belongs in smaller forums where the trade-offs can be debated honestly. An all-hands that runs tactical is a sign the org’s middle layer isn’t doing its job, not a feature.

#Four habits I had to build

The new ones don’t come naturally. They get practiced.

1. Ask the question that doesn’t have an answer yet. Operators get paid to ask the questions the team hasn’t thought to ask. Most of mine arrive in the form what would have to be true for this to break in six months? The question earns nothing in founder mode and earns its weight every quarter in operator mode.

2. Name the tradeoff in public. Founders make tradeoffs in their head and ship the result. Operators name what got chosen, what got declined, and what the consequence of being wrong looks like, in front of the team. The public-ness is what makes the decision reviewable — by the team, by the board, by future-me.

3. Defend the boring infrastructure work. Nobody on the team lobbies for the migration that pays back in two quarters. The lobbying job is the operator’s. Founders ship features. Operators also ship the conditions that let the team keep shipping features.

4. Hold the quality line when the team would ship. The hardest operator habit. Operators sometimes decide that something should be built better, even when the team is ready to ship it as-is. Holding that line looks like a brake. It is actually the spec.

The slower path through the team usually produces a worse answer once and a better team forever. The founder reflex hates this trade. The operator reflex depends on it.

#Both modes live in the same week

Most engineering leaders I respect run both modes inside a single week, not in separate phases of a career. A fifty-person org needs zero-to-one initiatives and selective pruning, often from the same person, sometimes the same day. This post treats the modes as serial because the shift is what’s hard to talk about. Inside any given role they live next to each other on the calendar. The challenge is rarely which mode to be in. The challenge is noticing which mode the room needs in this hour, and finding the reflex for that mode on demand.

#On “Founder Mode”

Paul Graham’s “Founder Mode” essay landed in 2024 and got broader treatment than it asked for. ¹ The essay names a real thing: founders run companies differently than career managers, and at some sizes the founder’s way produces strictly better results. The argument is precise on its narrow claim.

The argument this post makes is in tension with that one. Paul Graham’s essay reads as a permission slip to stay in founder mode longer than the career-manager class will tell you to. This post is the opposite permission slip. The same instincts that make a great zero-to-one CEO actively damage a 200-person org. Founder mode is a mode you can stay in too long. Operator mode is a mode some founders never learn to enter, and the company eats itself when they don’t.

The two arguments are both true. Founder mode is a useful, undervalued, non-default operating style at scale. Operator mode is a useful, underwritten, second-act operating style for the same person. The question isn’t which mode is better. The question is whether you can move between them when the company needs you to. Andy Grove’s High Output Management and Will Larson’s An Elegant Puzzle are both operator-mode books written without apology for the operator side of the work. ²

#What I’m still bad at

Holding the quality line, mostly. Letting the team ship while disagreeing with them out loud about the spec. The founder reflex says ship it, revise after. The operator reflex says the spec is the spec and we agreed. Both reflexes are useful. Knowing which one to run on a Tuesday afternoon is a skill I have not finished learning.

Also: operator mode as the only person in the room running it. Earlier in my career I worked alongside people whose dispositions covered the modes I wasn’t running. Operator mode in the abstract and operator mode when the team is calibrating to whether you’ll actually sit in the chair are different shapes of the same skill, and the second one only shows up when nobody else is doing it for you.

When I shift again I want to read this and find at least one claim I now think is wrong. That’s the test of whether the work was actual learning or just a clean story.

Notes

1. Paul Graham, "Founder Mode" (September 2024). ↩
2. Andy Grove, High Output Management (1983). Will Larson, An Elegant Puzzle (2019). Ben Horowitz, The Hard Thing About Hard Things (2014) -- particularly the peacetime CEO / wartime CEO chapter, which is the same observation through a different lens. ↩