The founder's bottleneck: which decisions to delegate, which never, and how to build judgment in your team
Business · · by Gabriel Bertagnolli
Series: Decisions that Move the Needle: Article 9 of 10
In any company that grows around a founder, the day comes when the bottleneck stops being the market, capital, or the product, and becomes a person. Decisions pile up in an inbox (the owner's), the team waits for approval on the smallest things, and the founder lives the paradox of working harder than ever while the company moves slower than ever.
It is not a problem of character or a generic "failure to delegate." It is a problem of decision architecture: which ones are centralized, which are delegated, and how the team's judgment is built so that delegation is not a leap in the dark. This article builds that architecture using the available evidence, from Noam Wasserman's data on founder survival in office, to Porter and Nohria's study on how CEOs use their time, and the one-way and two-way door principle popularized by Jeff Bezos at Amazon.
The evidence: the leader's time is the most misallocated resource
Let us start by sizing the resource. The most comprehensive study on executive time use is by Michael Porter and Nitin Nohria, which granularly tracked 60,000 hours of work across 27 CEOs of large companies for 13 weeks each (HBS / HBR, How CEOs Manage Time). One of its most cited findings: CEOs spend 72% of their working time in meetings, leaving only 28% for solitary work (CNBC on the study, 2018). In an SMB, where the infrastructure that absorbs the burden in large corporations does not exist, the concentration of decisions in one person is even more extreme.
The cost of that concentration becomes evident when the founder fails to delegate: the entire organization operates at the speed of a single calendar. And the opposite cost is also measured: when the founder delegates poorly or loses control over what is irrelevant, the result can be the founder's own departure. Wasserman's data is devastating in both directions: by the third year of ventures, 50% of founders were no longer CEO; by year four, only 40% remained in the seat, and by the time of an IPO, fewer than 25% remained (Wasserman, The Founder's Dilemma, HBR). His research on founding teams further concluded that around 65% of high-potential startup failures are attributed to co-founder and executive team conflicts, that is, people and control distribution issues, not product or market problems (Edison Partners on Wasserman's research).
And the delegation capacity of the average leadership does not help: in DDI assessments of more than 70,000 candidates for management positions, only 19% demonstrated strong delegation skills (DDI, 2025). The honest conclusion is twofold: delegating is structurally necessary, and almost no one does it well naturally. It is a competency that is designed, not a talent that is simply expected.
The organizing principle: one-way and two-way doors
Before the matrix, the principle that sustains it. Jeff Bezos formulated it in his 2016 Amazon shareholder letter: there are "one-way" decisions (consequential, irreversible or nearly so, requiring deep analysis and upper-level involvement) and "two-way" decisions, reversible, that can and should be made quickly by individuals or teams close to the problem, because if they make a mistake, you can walk back through the door (Bezos, 2016, Letter to Shareholders).
The most common organizational mistake is handling everything with the one-way decision process: committees, analysis, and approvals for what is reversible. The result is the slowness described by McKinsey in its research on decisions (more than 530,000 days of management time lost to decision inefficiency in an average Fortune 500 company, according to its survey, cited with full source in the first article of this series: McKinsey, 2019). The reverse error, recklessly making irreversible decisions, is rarer, but more expensive.
The delegation matrix: reversibility × impact
Cross the two variables that define the "door type": how reversible the decision is and how high its potential impact is. The quadrant where it falls defines who decides and through what process.
| Low potential impact | High potential impact | |
|---|---|---|
| Reversible (two-way door) | Fully delegate. The team decides alone; the founder finds out in the periodic review. Examples: discounts within band, weekly priorities, minor vendors, campaign formats. | Delegate with parameters. An owner with written criteria (limits, metrics, escalation conditions). The founder validates the framework once and reviews results, not decisions. Examples: pricing within ranges, operational-level hires, acquisition channel to test. |
| Irreversible (one-way door) | Decide quickly at lower levels, with a record. They do not warrant slowness: resolve in the weekly meeting with available information. Examples: firing a small toxic client, shutting down a marginal product line, campaign name. | Decide personally (or with the board), with a formal process. These are the decisions that are not delegated: partners, merger/sale, capital, executive-level hires, major strategic pivot, setting the decision culture. Here, slow analysis is an investment, not a cost. |
Four clarifications that make the matrix work in practice:
- The goal is not to delegate more: it is to delegate the right things. The "reversible" row concentrates most of the daily decision volume; freeing up that row is what restores the founder's time for the high-impact column, where their judgment is irreplaceable.
- Delegation is transferred with criteria, not without it. Every delegated decision needs its written parameter: permitted range, success metric, escalation condition ("if it exceeds X, you consult me"). Delegating without parameters is not autonomy: it is abandonment.
- Irreversible + low impact is decided quickly. The fear of making mistakes on small, irreversible decisions (shutting down a marginal product) is one of the engines of the bottleneck: they are not important, but they clog the inbox as if they were.
- The matrix is reviewed when scale changes. Decisions that were two-way with 10 employees can become one-way with 50 (a systems change affecting everyone). Review the classification once a year or upon jumping to a new revenue tier.
How to build judgment in the team (the part nobody does)
The most common reason founders do not delegate is not control: it is legitimate distrust that the team will make poor decisions. The answer is not to stop delegating, it is to build judgment deliberately. Three operational practices:
Practice 1: decisions with memory. Every important delegated decision is documented in a single line: context, options considered, selection criteria, expected outcome, and review date. Over time, that archive becomes the company's decision manual, the asset that allows delegating the next decision without starting from scratch.
Practice 2: review the process, not the decision. When a delegated decision goes wrong (and it will go wrong), the founder reviews with the owner how the decision was made: what information was considered, what options were weighed, what criteria were applied. If the process was sound, the mistake is accepted as a learning cost and the judgment becomes stronger. If the process was flawed, that is where the lesson lies. Punishing the outcome instead of the process destroys the willingness to decide.
Practice 3: the error budget. Explicitly state the acceptable margin of error: "out of every 10 decisions you make without consulting me, I expect 1 or 2 to go wrong; that is the cost of the system, and it is worth it." Without that explicit permission, the team escalates everything out of caution and the bottleneck reconstitutes itself. (McKinsey's evidence on decision velocity and quality also shows that deciding fast does not degrade quality; the fear of fast mistakes is, for the most part, a myth.)
Self-assessment: are you the bottleneck?
| Signal | Does this happen to you? |
|---|---|
| The team consults you on things they already decided last month | ☐ |
| Your approval inbox grows even though you work longer hours than anyone | ☐ |
| Vacations (yours or a key person's) paralyze processes | ☐ |
| No one on the team can answer "who decides this?" without mentioning you | ☐ |
| You decide things that you yourself consider minor | ☐ |
| Projects are delayed "waiting for the owner to find time" | ☐ |
Interpretation: 3 or more checked boxes indicate a structural bottleneck, not a workload issue. The answer is not managing your schedule better: it is redesigning who decides what, using the matrix above.
The final reframe
The founder is the only person in the company whose judgment is irreplaceable on one-way, high-impact decisions, and the only one whose attention is entirely replaceable in the rest. The bottleneck paradox is resolved when that asymmetry translates into architecture: the matrix decides what falls on you, written parameters transfer judgment, and decision memory accumulates the asset that allows the company to grow faster than your calendar. Wasserman's data serves as a reminder: remaining at the center of every decision does not preserve your control: it is the most common way to lose it.
Sources consulted
- Porter, M. E. & Nohria, N. (2018). How CEOs Manage Time (Harvard Business Review), faculty research page, Harvard Business School. Accessed October 14, 2026: https://www.hbs.edu/faculty/Pages/item.aspx?num=54691
- CNBC (2018). Harvard study: What CEOs do all day (June 20, 2018). Accessed October 14, 2026: https://www.cnbc.com/2018/06/20/harvard-study-what-ceos-do-all-day.html
- Wasserman, N. The Founder's Dilemma (Harvard Business Review, February 2008, article PDF). Accessed October 14, 2026: https://www.palominocap.com/wp-content/uploads/2017/06/5-HBR_Noam-Wasserman_The-Founders-Dilemma.pdf
- Edison Partners. What Every Founder Can Learn from Noam Wasserman. Accessed October 14, 2026: https://www.edisonpartners.com/blog/founders-dilemmas-business-building-with-professor-noam-wasserman
- Bezos, J. (2016). 2016 Letter to Shareholders (Amazon.com, published in 2018). Accessed October 14, 2026: https://www.aboutamazon.com/news/company-news/2016-letter-to-shareholders
- DDI (2025). DDI Data Reveals Delegation Is Top Factor in Preventing Burnout / Global Leadership Forecast. Accessed October 14, 2026: https://www.ddi.com/about/media/burnout
