पाठशाला Pathshala · दल Dal, The team · Lesson 27 · Scale
Replacing yourself: the roles a founder must give up
A company outgrows its founder one job at a time. Which jobs go first, how to write a function down so someone else can run it, and the delegation and numbers that stop the founder taking it back.
Pathshala, The Founder Library · 11 October 2026 · 7 min read
In its first year a founder is the company’s salesperson, product manager, recruiter, finance team and office manager. By the time it has a hundred people a founder who still holds those jobs is the reason it does not have two hundred. Replacing yourself is not one decision; it is a sequence of handovers, each of which has to be done properly or it is not done at all.
This lesson sets the order in which functions usually leave a founder’s hands, the manual that makes a function transferable, a three-stage handover over a quarter, the numbers that make trust cheap, and what stays with the founder. The [bottleneck lesson](/library/founder-as-the-companys-bottleneck) covers individual decisions; this one covers whole jobs.
Giving away a job every few months
Molly Graham puts it bluntly: if you want to grow as fast as your company, you have to give away your job every couple of months. She calls it giving away your Legos. The pieces a founder built feel like theirs, and the anxiety of watching someone else hold them is real and usually passes in a few weeks. Her advice for the person letting go is to find the next, larger problem rather than guard the old one. For a founder the next problem is always there: the company’s next stage, which nobody else is placed to work on.
The cost of not doing it is concrete. A founder who keeps a function does not just slow it down; they make the people in it into assistants, and the strong ones leave. They also make the next leader hire impossible, because no experienced leader will take a job the founder is still doing.
The order of the handover
Sort the founder’s jobs into four groups and hand them over in this order. First, work others can do as well as the founder: administration, scheduling, bookkeeping, vendor management, first-round interviews. These should go before the company has twenty people and rarely need a senior hire. Second, work the founder is weak at. For a technical founder this is often finance and operations; for a commercial founder, engineering management. Hire someone better than you, which is the easiest handover because the improvement shows at once.
Third, work the founder is good at but is no longer the only one who can do: founder-led sales once the [playbook](/library/sales-playbook-from-heroics-to-repeatability) is written and someone else has closed deals with it; product management once the product has more surface than one person can hold; recruiting once the hiring process runs without the founder in every loop. This is the painful group, because the founder is often still the best at it and the company notices a dip. Hand it over anyway; the dip is the price of the company being able to do the work twice as often.
Fourth, what stays. Paul Graham’s essay on founder mode warns against the conventional advice to hire good people and give them room, which he argues often means hiring professional fakers and letting them run the company into the ground. The answer is not to keep every job but to keep the right few: the company’s direction and its biggest bets, raising capital, choosing and running the top team, the culture, and the standard for the one craft the company exists for. Write that list down. It is short, and it is what the founder is for.
Indian companies add a fifth, quieter group: jobs tied to a founder by paperwork. The founder is often the only authorised signatory on the bank accounts, the director who signs the statutory filings, the name on the vendor contracts and the person every investor calls. Some of these should stay with a director; most of the work behind them should not. Add a second signatory with sensible limits, give the finance owner the filing calendar and the preparation, and keep only the signature. A founder who still prepares what they sign has not handed anything over.
Write the job down first
A function can be handed over only once it has been described. Log two weeks of the founder’s time in it: every meeting, task and decision. Then write the function’s manual on two pages: what it is for, the three to five numbers that say whether it is working, its weekly and monthly rhythm, the decisions it makes and who makes each, and the people outside the company who expect to deal with it. Mark the decisions the founder keeps and say why; everything else goes. Ben Horowitz’s advice on scaling applies here: it is much easier to add new people to old processes than new processes to old people, so write the rhythm down before the new owner arrives, not after.
A worked case. The founder of a 90-person software company in Bengaluru still runs sales on top of being CEO. The log shows twenty-two hours a week in it: twelve on deals above ₹25 lakh a year, six on the weekly pipeline review and forecast, four on pricing exceptions. The manual names four numbers: new annual contract value, win rate, sales cycle length and forecast accuracy. The founder keeps two things in writing: pricing exceptions above twenty per cent and the five largest accounts’ executive relationships. Everything else, including deals of any size, goes to the new head of sales. The founder’s twenty-two hours become about five.
Hand over in three stages
Handovers fail in two ways: the founder disappears on day one and the owner flounders, or the founder never quite leaves and the owner becomes a deputy. A quarter in three stages avoids both. In the first month the owner runs the function’s meetings with the founder present and mostly quiet. In the second the owner decides, and the founder checks large pieces of work at two points: Stripe’s practice, described by Krithika Shankarraman, is a review at about twenty per cent to agree the intent and another at eighty per cent, not ninety-nine, while there is still time to change course. In the third the owner decides and tells the founder afterwards, in a short weekly note.
Jay Desai’s line from the same piece states the logic: be hands-on until you trust someone. The stages are how trust is earned and shown. When something goes wrong in stage three, diagnose whether trust or skill has slipped before stepping back in, and then step back out. Work through one function at a time with the list below.
Keep the few things only a founder can do. Write everything else down, give it to someone by name, and stop doing it where the company can see.
Numbers that make trust cheap
A founder who has handed a function over still needs to know how it is doing. The worst way to find out is by asking around, which undermines the owner and teaches the team that the founder still decides. The best way is the function’s own numbers, on a dashboard the founder reads weekly, set before the handover and agreed with the owner. When a number moves, the founder asks the owner, not the team. The [weekly metrics lesson](/library/weekly-metrics-review-one-page-one-hour) has the format.
Numbers do not replace judgement. Mike Brown, quoted in the same First Round piece, describes porpoising: staying aware of everything at the surface and diving deep on a few initiatives each quarter. Paul Graham’s founder mode adds skip-level conversations as a normal practice rather than a sign of distrust. Both are ways for a founder to see a function closely without running it. The test is whether the founder comes back out: a dive that ends with the founder making the function’s decisions again is a handover reversed.
When the founder takes the job back
It happens, and it is usually gradual: one decision overruled, one meeting attended, one customer call taken directly. Each feels justified. Together they tell the owner the handover was not real. If the owner is not working out, say so and act on it, with the [leadership review](/library/building-leadership-team-first-vp) at day ninety; do not quietly do their job alongside them. And if the founder simply misses the work, which is common and human, the [identity shift lesson](/library/from-builder-to-manager-founders-identity-shift) is about exactly that.
The quarterly handover review
Once a quarter, the founder spends an afternoon on three questions. Where did my hours go, by function, from the calendar? Which of those functions now has, or should have, an owner, and which group does each fall in? And for every function handed over in the last two quarters, have I stopped doing it, and is it better, the same or worse? Pick the next one function to hand over and start its two-week log on Monday. Share the answers with the leadership team; they will notice the founder’s backsliding before the founder does.
The order is a default, not a law: a founder’s strengths and the company’s constraint decide the exceptions.
Sources
- First Round Review, Molly Graham: Give Away Your Legos and Other Commandments for Scaling Startups
- Paul Graham, Founder Mode, September 2024
- First Round Review, Is All Micromanagement Bad? Here’s How the Best Startup Leaders Balance Details and Delegation, 11 June 2025 — Stripe’s 20 and 80 per cent checkpoints; Jay Desai on trust; Mike Brown on porpoising.
- Ben Horowitz, Taking the Mystery Out of Scaling a Company, a16z