पाठशाला Pathshala · मन Man, The founder · Lesson 11 · Build
The founder as the company’s bottleneck
At eight people the founder deciding everything is a strength. At fifteen it is a queue. Find the decisions only you are making, keep the few that need you and hand the rest over within a quarter.
Pathshala, The Founder Library · 11 October 2026 · 7 min read
At eight people, a founder who sees every decision is why the company moves fast. At fifteen the same habit is why it has stopped. Nothing about the founder changed. The arithmetic did, and it can be computed.
This lesson treats the founder as what the company has quietly made them: a single point through which work must pass. It explains why that point jams suddenly rather than gradually, gives a two-week audit that finds the decisions only the founder is making, and sets out how to hand the unnecessary ones over within a quarter without losing the standards that made the company work.
How to tell you are the bottleneck
The symptoms appear before anyone names them. Work waits on you. Ask the team what they are blocked on and listen for your own name. Your inbox is a queue. Messages that start with “quick question” or “can you approve” arrive faster than you clear them, and the oldest is from last week. Decisions reopen. A call made by a lead is reversed when you see it, so the team learns to wait for you rather than decide. You are in every meeting because nothing moves without you, and you are behind on everything because you are in every meeting. The team’s best people are quietly frustrated, because they were hired to decide and are being asked to propose.
None of these is a character flaw. Each is the natural outcome of a founder doing exactly the right thing at five people and continuing to do it at fifteen. The difficulty is that the transition is not felt as it happens. A week is a little busier than the last, the next a little more. Then a quarter has gone and the company is slower than it was at half the size.
The arithmetic of a queue
A founder who must answer every approval is, in the language of operations, a single server. Decisions arrive at some rate, set mainly by the number of people in the company and how many things each of them needs from the founder. The founder clears them at some rate, set by the hours available and the minutes each decision takes. Queueing theory has a well-known and unforgiving result for this arrangement: the average time a decision spends waiting and being handled is one divided by the gap between capacity and arrivals.
The consequence is not intuitive. When the founder is half busy, the wait is about twice the time the decision takes. At four-fifths busy it is five times. At nine-tenths it is ten times. As arrivals approach capacity, the wait grows without limit, and a single extra hire can tip a company from a wait of a day to a wait of a week. That is why the jam feels sudden. It is not that the founder got worse. The company added three people, and the queue crossed the line.
Try the obvious fixes and watch which one works. Add founder hours and the line moves right a little, at the cost of everything else the founder should be doing. Cut minutes per decision and it moves a little more. Cut the decisions each person brings and it moves a great deal. The lever that matters is the number of decisions that need the founder at all, and that lever is pulled by delegation and elimination, not effort.
The decision audit: two weeks, every decision
For ten working days, write down every decision you make, however small, in a sheet with five columns. What it was. Who brought it. Minutes it took, including the switch from what you were doing. Reversible? Yes, mostly or no, using the test in the [lesson on reversible and irreversible decisions](/library/decisions-reversible-irreversible-how-fast). Could someone else have made it as well as you, or well enough? Founders who do this honestly are usually surprised less by the total than by the shape: a long tail of small approvals they did not know they were making.
Then sort every line into four boxes. Only you: the one-way doors and the things that set direction, such as raising money, senior hires, the company’s values in practice, a strategic customer, a price change that resets the market. Delegate with a checkpoint: important but reversible decisions where someone else can decide if you see the plan early. Delegate fully: reversible, routine, and better made by the person closest to it. Eliminate: decisions that should not exist at all, such as approvals nobody remembers asking for. Sam Corcos, describing how he works with assistants in First Round Review, puts the order of operations well: never automate something that can be eliminated, and never delegate something that can be automated or streamlined. Eliminate first.
The lever that matters is not the founder’s effort. It is the number of decisions that need the founder at all.
Handing over within a quarter
A thirteen-week plan works for most companies between ten and forty people. Weeks one and two: the audit. Week three: sort and choose. Pick the ten most frequent decisions in the delegate boxes and name an owner for each. Week four: write a decision right for each, one line long, in a shared document: who decides, within what limits, who must be told. “Customer refunds up to ₹25,000: support lead decides, tells finance weekly.” “Freelancer contracts under ₹2 lakh a month: the hiring manager decides, tells the founder after.” “Travel within the policy: the traveller books, finance checks monthly.” Limits make delegation safe; a right without a limit is either abdicated or ignored. Publish the document where the whole team can read it, so people stop bringing you what is no longer yours to decide. Weeks five to ten: run it. Weeks eleven to thirteen: review, widen the limits that worked, take back the few that did not.
Two practices make the handover stick. The first is checking early rather than late. First Round’s review of how founders balance detail and delegation describes review checkpoints at about 20 per cent, when the strategy can still change, and 80 per cent, when execution can still be corrected. A founder who sees only the finished work reverses it and teaches the team to wait; one who sees the plan early can shape it and then let go. The same piece quotes Jay Desai: hands-on until trust is established, then hands-off. The second is budgeting for wrong calls. Delegated decisions will sometimes be worse than the founder’s would have been. If every wrong call is reversed in public, delegation ends. Fix the pattern privately; leave the individual decision alone unless it locks a door.
What stays with the founder
Delegation is not withdrawal. Paul Graham’s Founder Mode argues that the standard advice to hire good people and give them room has let some companies drift, and that founders who stay close to the work, including through skip-level conversations, often run their companies better than the advice predicts. The lesson here is consistent with that. The founder should stay close to the work. What the founder should stop doing is approving it.
The distinction is between attention and permission. A founder can read support tickets, sit in on sales calls and talk to engineers three levels down without becoming the gate anything must pass through. Jeff Bezos, in his 2016 letter to shareholders, made the related point that most decisions are reversible two-way doors that can use a lightweight process, and that most should be made with around 70 per cent of the information one wishes one had. Keep the one-way doors. Watch everything. Approve very little.
A quarterly rerun, in one afternoon
In the first week of every quarter, log decisions for three days rather than ten, and compare with the last audit. Count how many still need you and whether that number grew with headcount. Reread the decision rights document and widen at least two limits that held well. Take back any right that produced a pattern of wrong calls and write down why. Then rerun the figure above with the coming quarter’s hiring plan: if the line shows the company reaching your capacity before the quarter ends, the next round of delegation starts now, not when the queue is already a week long. The founder’s job at every size is to remain the person the company needs least for the things that can be done without them, and most for the few that cannot.
The figure is a model, not a measurement. Real decisions arrive in bursts and vary in size, which makes real queues longer than the curve suggests. Sources were checked in October 2026.
Sources
- First Round Review, Founders, Use This Tactical Guide for Working with EAs to Get Better at Delegating, May 2022 — Sam Corcos: eliminate, then automate, then delegate.
- First Round Review, Is All Micromanagement Bad? Here’s How the Best Startup Leaders Balance Details and Delegation, June 2025 — Checkpoints at 20 and 80 per cent; Jay Desai on trust.
- Paul Graham, Founder Mode, September 2024
- Jeff Bezos, 2016 Letter to Shareholders, Amazon — Two-way doors; decisions at around 70 per cent of the information.