How do you coach a startup?

You coach a startup by coaching the people running it, because at that size the company is largely a reflection of a few individuals' judgment. The work changes sharply by stage: early on it is focus and founder honesty, in the middle it is delegation and hiring, later it is the founder removing themselves from the critical path.

Coaching a startup like a corporate leadership programme is the most common way this goes wrong.

What is different about startups

Three things that change how the work has to run.

Speed of change. A plan made in a session can be obsolete in ten days. Long structured programmes with fixed modules do not survive contact with a company that pivots twice a year. The work has to start from whatever is actually happening this fortnight.

Concentration of decisions. In a fifty-person company, a handful of people make almost everything that matters. Improve one founder's judgment and you have moved the whole organisation, which is a kind of reach you do not get in a large company.

Existential stakes. A corporate executive making a poor call has a bad year. A founder making a poor call can end the company and everyone's job with it. That changes how much time you spend testing a decision before it gets made.

Stage one: focus and honesty

Pre-revenue to early traction. Small team, no structure, everything urgent.

The work here is almost entirely about focus. Founders at this stage are doing eleven things, six of which do not matter, and they cannot tell which six. Good coaching is relentless about what gets dropped.

The second theme is honesty about traction. Early founders are skilled at reading signal into noise. Three friendly calls become a market. A pilot that never converted becomes a case study. A useful coach asks what would have to be true for this to be real, and what evidence would change your mind.

Many companies at this stage do not need a coach, which is the argument in do entrepreneurs need a coach. If you are pre-product with two people and no revenue, another month of building beats almost any session. I turn this work away.

Stage two: delegation and hiring

Roughly twenty to eighty people. The stage where most founder tenures come apart.

The job has changed and the founder has not. They were good at making the thing, and now the job is making the group that makes the thing. These are different skills, and nobody announces the switch.

The work here is concrete. Which decisions still route through you and why. What happened the last two times you delegated something and took it back. Who on the leadership team you hired for a company that no longer exists, and what you are avoiding by not addressing it.

One founder at this stage had handed a major partnership to a colleague, then invited himself to their first solo meeting and overruled them in front of the partner. He was probably right on the substance. He also taught the partner who really decides and taught the colleague what handovers are worth, and found himself running the relationship again within a month. That is the stage in one scene.

The co-founder relationship also becomes live at this stage. Companies rarely die of strategy. They die of two people who stopped talking honestly eighteen months ago and quietly built an organisation around the gap. Meetings scheduled to avoid a conversation. Teams split so two people never have to agree. A co-founder alignment sprint is the fastest way I know to interrupt that.

Stage three: removing yourself

Roughly a hundred people and up. The founder is now the single largest constraint on the business and often the last to know.

The work becomes about what only they can do: deciding what the company will not do, setting the standard by what they tolerate, choosing the people who choose the people, and telling the truth about the situation especially when it is bad. Everything else is a candidate for someone better at it.

This is where the work gets personal, and not by choice. It is the stage the purpose of a founder is really about. Removing yourself means dismantling the daily proof that you matter. Founders delegate, watch it get done differently, feel it, and take the work back. Twice, usually, before anyone names the loop.

What the sessions actually look like

Regardless of stage, the mechanics in my practice are the same.

Sessions every two weeks, sixty to ninety minutes. The first ten minutes are clearing, and what comes out is messier than in any other meeting because there is no reason to tidy it. Then one or two things get taken seriously rather than eight things getting touched. The last stretch is commitment: something specific, by a specific date.

Between sessions you write. Two or three paragraphs on what you committed to and what actually happened. Half of my clients find their real position while writing rather than while talking.

Engagements run six to twelve months. Shorter and you get insight without change.

Coaching a founding team rather than one person

Sometimes the right unit is two or three people, not one. Co-founder work is a different discipline and worth naming.

It needs separate individual sessions as well as joint ones, because the honest material rarely arrives first in a room with the other person in it. It needs an explicit agreement about what carries between sessions. And it needs both people to want it, because one founder dragging another into the work produces theatre.

I have run this under real time pressure. Two founders, eight years in, had not spoken for two months by the time they called me. Weekly sessions with each of them alone, weekly sessions with both, and we started by writing down what each of them was actually responsible for and how decisions got made between them. Then each wrote the other a contract: here is what you need to do, otherwise I cannot continue. It took several rounds before either could sign anything. It worked, and what made it work was not the document. It was that the process made both of them notice they still wanted to build the thing together.

The most useful hour two co-founders can spend is an unromantic conversation about what happens if one of them wants out. It gets avoided because it feels like planning for failure. It is far cheaper before anything is wrong.

Common questions

Can you coach a whole team at once?
You can run team sessions, and they solve different problems. The judgment work happens one to one. Group settings surface issues and rarely resolve the personal ones underneath.

How is this different from an accelerator?
Accelerators deliver curriculum, network and money to a cohort. Coaching is one person's situation over months. They complement each other and neither replaces the other.

What if the founder is the problem?
Frequently they are, and everyone knows it except them. That is what the work is for. It requires a coach with no stake in the answer and a willingness to be unpopular.

When is it too early?
Pre-product, two people, no revenue. Ship first. See do entrepreneurs need a coach.

The short version

Coach the people, not the company, and match the work to the stage. Focus early, delegation in the middle, removing yourself later. Anything that ignores the stage is a programme rather than coaching.

More on the category in what is startup coaching.