Founder Decision-Making Framework: How to Make Hard Calls
A practical framework for founder decision-making — covering reversible vs irreversible calls, pre-mortems, decision journals, and beating cognitive bias.
Whether you're looking for an angel investor, a growth advisor, or just want to connect — I'm always open to great ideas.
Get in TouchAI, startups & growth insights. No spam.
TL;DR: At every stage of a startup, the quality of founder decision-making is the single most constraining variable in the company's trajectory. Not capital. Not market timing. Not team. Decisions. This post gives you a complete operating system: how to classify decisions by type, the right process for each class, OODA loop for fast operational calls, pre-mortem for high-stakes strategic ones, a framework for designing decision authority across solo, team, and board contexts, the disagree-and-commit protocol, how to move with incomplete information, and a decision journal template you can start using today.
There is a specific moment that most founders recognize in retrospect but rarely see clearly in the moment. It is the moment when the company's growth, culture, or survival stops being limited by market conditions and starts being limited by the founder's decision-making quality and velocity.
I have seen this from both sides — as a founder and as an investor watching portfolio companies. When a company is stuck, the diagnosis often sounds like a product problem or a hiring problem or a GTM problem. But when you go one level deeper, you almost always find that the root cause is a decision that was deferred too long, made by the wrong person, made with the wrong process, or made and then reversed for the wrong reasons.
The decision-making bottleneck takes different forms at different stages. The Founder Challenges Checklist maps the most common structural mistakes across Years 1-3 that compound this bottleneck — reading it alongside this framework gives you both the diagnostic and the operating system.
Pre-product-market fit: Founders make too many decisions by committee or research everything exhaustively in a domain where speed of iteration is the only thing that matters. The cost of deliberation vastly exceeds the cost of being wrong and correcting. Knowing when you have actually achieved product-market fit — rather than assuming it — is itself one of the most consequential decisions founders routinely get wrong.
Post-PMF, pre-Series A: Founders stop making decisions because the organization hasn't been designed to make them. Decisions drift toward whoever speaks loudest in a room. Accountability diffuses. Things that need a clear answer get tabled or re-opened.
Post-Series A: Founders make decisions without adequate input from the team that has to execute them. Or they abdicate decisions to the board that should be made at the operator level. Or they make decisions and then quietly second-guess them in ways that undermine their own credibility.
At scale: The founder becomes a bottleneck because every significant decision still flows through them. The organization hasn't been designed to push decisions down, and the founder hasn't built the judgment in their team to support that.
In every one of these failure modes, the problem is not that the founder lacked information or intelligence. The problem is that they lacked a system — a structured approach to classifying decisions, designing the process for each type, and building an organization that can execute at the right decision velocity.
That system is what this post gives you.
The best founders I know do not make better individual decisions than everyone else. They have better decision processes — and they have internalized those processes deeply enough to apply them instinctively under pressure.
Before you can choose the right process for a decision, you need to classify it. Most founders treat all decisions similarly — they use roughly the same amount of mental energy, deliberation time, and stakeholder involvement regardless of whether they are choosing a new office snack policy or deciding to pivot the company.
That is a catastrophic allocation error. A decision that deserves 15 minutes gets four weeks. A decision that deserves four weeks gets 15 minutes. The result is organizational whiplash, founder exhaustion, and a team that does not know what to prioritize.
Here is the classification system I use. Every decision a founder faces falls into one of three primary categories across two dimensions.
| Type | Definition | Example | Default Process |
|---|---|---|---|
| Type 1 (Irreversible) | One-way doors. Cannot be undone without significant cost, time, or reputational damage. | Pivoting the core product, choosing a co-investor, firing an executive | Slow down. Get more data. Pressure-test hard. |
| Type 2 (Reversible) | Two-way doors. Can be undone quickly at low cost. | Changing a pricing page, testing a new acquisition channel, restructuring a team meeting | Speed up. Default to action. Correct on the fly. |
Jeff Bezos introduced this framework in his 2015 Amazon shareholder letter and it remains one of the most useful mental models in startup leadership. The error most founders make is treating Type 2 decisions like Type 1 — applying the full weight of deliberation to something that should be a 48-hour test.
The reverse error is less common but more catastrophic: treating a Type 1 decision like a Type 2. This is how founders make board composition decisions based on momentum, or choose architectural paths for the product without thinking through the long-term lock-in, or let people go without adequate documentation — and then discover the decision has an enormous tail of consequences they didn't anticipate.
The first question to ask about any decision is: if I am wrong, how do I undo this? If the answer is "easily, within a week, at low cost," it is a Type 2. If the answer is "I can't, or it would take years and enormous resources," it is a Type 1.
The second dimension is domain. Decisions in different domains have different information requirements, different stakeholders, and different failure modes.
| Domain | Characteristics | Common Pitfall |
|---|---|---|
| People | High irreversibility, high emotional load, reputational stakes, legal dimension | Moving too slowly on underperformance; applying subjective criteria inconsistently |
| Product | Variable reversibility, high iteration value, strong data signal available | Over-weighting opinion and intuition vs. user behavior data |
| Process | Usually reversible, high leverage on team efficiency, resistance to change | Designing for current team size rather than the team you will have in 12 months |
When you overlay these two dimensions, you get a 2x3 matrix. The cells at the intersection of "irreversible" and "people" deserve the highest deliberation cost. The cells at the intersection of "reversible" and "process" deserve the lowest.
In practice, many founder decisions are compound — they have elements of multiple categories. A decision to hire a new VP of Sales is a people decision (high irreversibility), a product decision (the GTM architecture changes), and a process decision (how the team works changes). For compound decisions, default to the process appropriate for the highest-stakes component.
The reversibility framework deserves its own treatment because founders consistently misclassify decisions on this dimension. Here is a more precise tool.
| Reversal Cost | Reversal Time | Classification | Recommended process speed |
|---|---|---|---|
| Low ($) | Fast (< 1 week) | Clear Type 2 | Hours to days |
| Low ($) | Slow (> 1 month) | Type 2 with friction | Days to weeks |
| High ($$) | Fast (< 1 week) | Financial Type 1 | Weeks |
| High ($$) | Slow (> 1 month) | True Type 1 | Weeks to months |
Some decisions that feel irreversible are actually Type 2 with high perceived reversal cost but low actual reversal cost. Changing a pricing model feels irreversible because it creates customer communication overhead. But in practice, SaaS pricing changes are made and reversed regularly without catastrophic consequence. The perceived cost — managing customer communication, potential churn risk — is real but manageable. Do not let perceived cost masquerade as true irreversibility.
Some decisions that feel reversible are actually Type 1 with delayed consequences. Culture decisions are the clearest example. Deciding to let a high-performer stay who is visibly undermining team dynamics feels reversible — you can always address the culture problem later. In practice, this decision compounds. The toxic behavior signals that it is tolerated. Others adapt to it. The culture calcifies. By the time you act, you are not reversing one decision; you are unwinding 18 months of organizational drift.
The most dangerous decisions are the ones that feel like Type 2 but have Type 1 consequences. Classify by actual reversal cost, not by perceived reversal effort.
For Type 2 decisions — the operational, reversible, day-to-day calls that consume the majority of a founder's decision bandwidth — the OODA loop is the right framework. Developed by military strategist John Boyd for fighter pilot decision cycles, it has been widely adopted in startup and business contexts.
OODA stands for Observe, Orient, Decide, Act.
Gather raw data without interpretation. In a startup context this means: customer support tickets, churn data, NPS scores, conversion metrics, team signals. The goal at the Observe stage is breadth and accuracy, not analysis. What is actually happening? What are you seeing in the data, in direct customer conversations, in how the team is behaving?
The observe stage fails when founders filter too aggressively before the data reaches them. If you only hear about problems that your direct reports have already categorized and prioritized, you are operating on a curated version of reality. Build direct observation channels — talk to customers directly, read raw support tickets, look at the underlying data before the dashboard.
This is the most important stage of the OODA loop, and the one most often skipped. Orient means contextualizing the observed data through your mental models, past experience, cultural context, and analytical frameworks.
Boyd's key insight was that orientation is not just analysis — it is the lens through which you interpret everything. If your mental model is wrong, correct data will produce wrong conclusions. This is why two equally intelligent founders looking at the same data can reach opposite conclusions: they are orienting through different lenses.
In practice, Orient means: What patterns does this data fit? What hypotheses does it generate? What has worked or failed in analogous situations? What are the first-order and second-order implications?
From your oriented view, select a course of action. For Type 2 operational decisions, the goal is to decide quickly with sufficient information — not to decide with complete information (which is never available) and not to decide prematurely (before the orientation stage has generated enough signal).
The decision should be specific and time-bounded. "We will test X for two weeks, measure Y, and adjust at the end of the two-week window" is a decision. "We should probably explore some options around X" is not.
Execute the decision fast and fully. A slow or half-committed execution degrades the quality of the data you get back, because you cannot tell whether the outcome reflects the quality of the decision or the quality of the execution.
After execution, the loop begins again. The new data from the action feeds back into Observe, and the cycle tightens over time.
The OODA advantage in startups is cycle time. The goal is not to make a perfect decision on each iteration — it is to cycle faster than the environment changes. A startup that can complete a full OODA loop in 48 hours will out-learn and out-adapt a competitor that takes two weeks, regardless of which has the better initial mental model.
| OODA Stage | Common failure mode | Fix |
|---|---|---|
| Observe | Data filtered before it reaches you | Build direct observation channels |
| Orient | Skipping to decide without contextualizing | Force a hypothesis step before any decision |
| Decide | Deciding before orientation is complete | Separate "gathering" meetings from "deciding" meetings |
| Act | Slow or uncommitted execution that muddies the signal | Full commitment with a defined review checkpoint |
For Type 1 decisions — the irreversible strategic calls — the OODA loop is insufficient. You need a process that surfaces the failure modes of a decision before you commit to it. The pre-mortem, developed by psychologist Gary Klein, is the most effective tool I know for this.
A standard post-mortem asks: now that this has failed, what went wrong? A pre-mortem reverses the time direction: assume that the decision you are about to make has been fully executed and has failed catastrophically. It is now 18 months in the future. What went wrong?
The power of this inversion is psychological. When you are in the process of deciding on a course of action, the natural human tendency is to look for confirming evidence and discount disconfirming evidence — a phenomenon called motivated reasoning. The pre-mortem short-circuits this by giving participants explicit permission to find every possible way the decision could fail.
State the decision clearly. Write it down in one precise sentence. "We are deciding to pivot from B2C to B2B SMB, targeting companies with 10–50 employees, with a 90-day timeline to complete the transition."
Set the failure assumption. "It is now 18 months from today. This pivot has failed completely. The company has lost significant revenue, the team is demoralized, and we are out of runway. This is not a hypothetical — it has definitively happened."
Individual brainstorm first. Give each person in the room 10 minutes to write down every specific reason the decision failed. No cross-talk. The individual-first format prevents the first person's answer from anchoring everyone else.
Read out and compile. Each person reads their list. Compile without judgment. The goal is a complete failure map, not an edited or prioritized one.
Cluster and prioritize. Group the failure modes by category. Identify the three to five failure modes that are both most probable and most catastrophic if they occur.
Mitigate or decide. For each high-priority failure mode, decide: Can we structurally mitigate this risk? If yes, build the mitigation into the decision. If no — if the failure mode is both probable and unmitigable — revisit whether to make the decision at all.
The pre-mortem does not guarantee good decisions. What it does is strip away the false confidence that comes from motivated reasoning and surface the failure modes that people are already privately worried about but are not saying out loud. In my experience, the most valuable pre-mortem output is almost always something someone in the room already knew and hadn't said because they didn't want to be the person who killed the plan.
Run a pre-mortem on every Type 1 decision. Budget at least 90 minutes. The ROI on that 90 minutes, relative to the cost of the decision going wrong, is always positive.
Who should make which decisions is one of the most important organizational design questions a founder faces, and most founding teams answer it by default rather than by design. The result is either decision centralization (everything flows to the founder, who becomes a bottleneck) or decision diffusion (nobody is clearly accountable, and things get resolved by whoever has the most social capital in the room).
Solo decisions are made by one person with explicit authority and accountability. They may be informed by input from others, but the final call belongs to one person who owns the outcome. The appropriate scope of solo decisions narrows as the company scales — but in the early stage (sub-10 employees), most decisions should be solo because the organizational overhead of group decision-making is too high.
Team decisions are made by a defined group through a structured process. They are appropriate for decisions with significant cross-functional implications — where the decision genuinely requires the expertise of multiple domains to make well, and where buy-in and alignment are as important as the quality of the decision itself.
Board decisions are made by the board with or without management recommendation. They are appropriate for decisions that materially affect investor economics or governance — major financing events, CEO compensation, significant M&A, strategic pivots that materially change the risk profile of the company.
Most founders make the mistake of escalating operational decisions to the board (seeking validation they don't need) and making board-level decisions unilaterally (because they haven't thought through governance). Both are expensive errors.
A simple but powerful tool is the RACI-adjacent decision matrix. For every category of decision your company faces, explicitly assign:
| Decision Category | Decider | Consulted | Informed |
|---|---|---|---|
| Product roadmap priority | CPO / Founder | Head of Sales, Head of Customer Success | Whole company |
| Individual compensation changes | CEO | Functional Head, CFO | Finance only |
| Fundraising term acceptance | CEO | Board, CFO | Whole company after close |
| Hiring a senior IC | Functional Head | CEO | Relevant team |
| Brand and messaging changes | CMO / Founder | CEO, Sales | Whole company |
| Engineering architecture choices | CTO | Relevant engineering leads | Engineering team |
Building this matrix explicitly — and then enforcing it consistently — does several things that are valuable. It frees people to make decisions without seeking permission they don't need. It prevents decisions from being relitigated after they are made. It forces clarity about who owns what before conflicts arise. And it signals to the team that decision authority is a deliberate organizational design, not an accident of whoever happened to be in the room. An advisory board built with this design intent — where each advisor covers a specific decision domain — amplifies this effect by giving you external judgment you can pull into the right tier without creating governance confusion.
One of the most corrosive patterns in startup decision-making is the silent dissenter — the person in the room who disagrees with a decision, does not say so clearly, and then either half-commits to execution or actively undermines it.
This pattern is so common because the social dynamics of startup teams make disagreement costly. Founders who push back on their co-founder's ideas are seen as difficult. Employees who disagree with a decision feel they are risking their standing. So people say yes in the meeting and communicate their doubts through inaction, passive resistance, or quiet negative framing to their peers.
The "disagree and commit" protocol is Amazon's solution to this problem, and it is one of the most operationally important cultural norms a founder can establish.
Step 1: Create a genuine window for dissent. Before a decision is finalized, the Decider explicitly opens the floor for disagreement. Not "does anyone have any thoughts?" but "I want to hear the strongest case against this decision. What are we missing? Who thinks this is wrong?" The explicit invitation changes the social dynamic.
Step 2: Document the disagreement. When someone disagrees, have them state their position clearly. Write it down. "Maria believes this pricing change will disproportionately hurt our mid-market segment and create a churn risk we haven't modeled." The documentation matters because it validates the dissent as real — it was heard, it is on record — which makes the next step possible.
Step 3: Make the decision. The Decider makes the call, taking the documented disagreement into account. They are not required to change their position. They are required to acknowledge the disagreement explicitly and explain why they are deciding as they are despite it.
Step 4: Commit fully. Once the decision is made, everyone — including the people who disagreed — commits fully to execution. The disagreement does not persist as a partial commitment. "I think this is wrong, and I will execute it 100%" is the expected and respected posture.
Step 5: Create a review checkpoint. Build in a defined moment to revisit the decision with actual data. "We will run this pricing change for 60 days and then review the churn data together." This gives the dissenters a legitimate forum for their concerns without undermining current execution.
The key to making disagree-and-commit work is that it protects both the Decider and the dissenter. The Decider gets clean execution without passive resistance. The dissenter gets their position documented and a review checkpoint where it can be validated. Neither party has to win or lose in the moment.
Disagree and commit is not about suppressing dissent. It is about giving dissent a structured home so it doesn't bleed into execution.
Founders consistently underestimate how much uncertainty is irreducible — how much of the information they would need to make a truly confident decision simply does not exist and cannot be obtained before a decision window closes.
This creates a specific failure mode: the perpetual deferral. The founder keeps waiting for more data, more signal, more clarity. The decision that needed to be made in Q1 gets made (badly, under pressure) in Q4. Or it never gets made at all, and the window closes.
Here is a practical framework for deciding when you have enough information to decide.
Jeff Bezos articulates this as the 70% rule: make most decisions when you have about 70% of the information you wish you had. If you wait for 90% certainty, you are moving too slow. If you consistently decide with less than 70%, you are being reckless.
The 70% threshold is calibrated by decision type. For a Type 2 reversible decision, 50% is often enough because the cost of being wrong is low. For a Type 1 irreversible decision, 70% might not be enough — you want to push toward higher confidence and use the pre-mortem to map the remaining 30% of uncertainty.
Before making any significant decision, answer three questions explicitly:
What would change my decision? This identifies the information that actually matters. If you cannot name a specific data point or finding that would cause you to decide differently, you are not information-limited — you have already decided and are rationalizing.
Can I get that information in a reasonable timeframe? If yes, defer briefly to gather it. If no — if the information genuinely does not exist or cannot be obtained before the window closes — recognize that you are deciding under irreducible uncertainty and adjust your process accordingly.
What is the cost of deferring vs. the cost of deciding wrong? This is the decision calculus that most founders skip. Deferral has a cost: team uncertainty, lost competitive window, organizational signal. Deciding wrong has a cost: the reversal cost for Type 2, or the structural damage for Type 1. Explicitly comparing these costs usually makes clear that one direction is more expensive than the other.
One of the highest-leverage behaviors a founder can develop is the ability to communicate decision confidence accurately. Founders who communicate "I am 90% confident in this" when they are actually 55% confident create organizational trust misalignments. The team calibrates their own confidence based on the founder's stated confidence. When the decision turns out to be wrong at a rate inconsistent with the expressed certainty, trust degrades.
The alternative is explicit confidence communication. "I am making this decision with about 60% confidence. Here is what I know, here is what I don't know, here is what would make me more confident, and here is why I am deciding now rather than waiting." This is not weakness — it is epistemic honesty, and it builds the kind of trust where teams feel safe surfacing dissent and new information.
The most systematic way to improve your decision-making over time is to maintain a decision journal — a structured record of the significant decisions you make, the reasoning behind them, and the outcomes.
Most founders do not do this because they feel like they don't have time and because reviewing past decisions is uncomfortable. Both of these resistances are exactly why the journal is valuable. It forces you to slow down at decision-making moments, and reviewing it surfaces the biases and patterns you would otherwise never see in yourself.
For every significant decision (Type 1 always; Type 2 when they are large or recurring), record:
| Field | What to write |
|---|---|
| Decision | The specific call being made, in one sentence |
| Date | When the decision was made |
| Type | Reversible / Irreversible; domain (people / product / process) |
| Confidence | Your confidence level at time of decision (0–100%) |
| Key information | The three most important facts or data points informing the decision |
| Key uncertainties | The two or three most significant things you do not know |
| Alternative considered | The strongest alternative option you rejected, and why |
| Expected outcome | What you expect to happen over what timeframe |
| Review date | When you will look back at this decision |
| Actual outcome | Filled in at review: what actually happened |
| Learning | What the gap between expected and actual tells you |
A decision journal without a review cadence is just a diary. The value comes from the review. I recommend:
The annual review in particular is uncomfortable in exactly the right way. You will find decisions you made with high confidence that turned out wrong. You will find decisions you made with low confidence that turned out right. The calibration between these two will tell you more about your actual decision quality than anything else.
Every founder is a human being, which means every founder is running on cognitive hardware that evolved for a very different environment than "deciding whether to pivot a SaaS company." The biases that this hardware produces are well-documented. The question is not whether you have them — you do — but whether you have designed your decision process to catch them.
Here are the ten most consequential cognitive biases for founders, and the practical antidotes.
| Bias | How it shows up | Antidote |
|---|---|---|
| Confirmation bias | You seek information that confirms the direction you're already leaning and discount disconfirming evidence | Pre-mortem. Steel-man the opposing view. Assign someone to argue against the decision. |
| Sunk cost fallacy | You continue investing in a direction because of what you've already spent, not because of future returns | Frame every decision as "if we were starting fresh today, would we invest in this?" |
| Availability heuristic | You overweight recent, vivid examples (the last customer call, the most recent hire who failed) when making general decisions | Require that every significant decision cite base rates and population-level data, not just recent examples |
| Overconfidence | You consistently underestimate uncertainty and overestimate your ability to predict outcomes | Track calibration via decision journal. Explicitly state confidence percentages. |
| Status quo bias | You overweight the costs of change and underweight the costs of staying the same | For every decision to maintain the status quo, require an explicit justification that acknowledges opportunity cost |
| Anchoring | The first number or option presented becomes the implicit reference point that distorts all subsequent evaluation | Generate your own estimate before seeing others'. In negotiations, set anchors deliberately and consciously. |
| Recency bias | Recent data dominates your view; older data patterns are discounted | Require that decisions include historical data going back at least 12 months, not just the last 30 days |
| Affect heuristic | Decisions are driven by emotional valence rather than analytical assessment — you like a founder, so you overweight their pitch | Separate "impressions" from "analysis" explicitly. Do your analytical work before meeting the person. |
| Planning fallacy | Projects take longer and cost more than estimated; benefits are smaller than projected | Use reference class forecasting: how long do comparable projects actually take, not how long does this one feel like it will take |
| Groupthink | Teams converge prematurely on a decision to maintain social harmony; dissenting views get suppressed | Require anonymous input before group discussion on significant decisions. Assign an explicit devil's advocate role. |
The metacognitive move here is important: knowing about these biases does not protect you from them. Research consistently shows that people who are most aware of cognitive biases are only slightly better calibrated than those who are not, because the biases operate below the level of conscious reasoning. The protection comes not from awareness but from structural process design — building in steps that catch biases mechanically, regardless of whether you are in the moment of experiencing them.
The right decision framework is not static — it evolves as the company scales.
0–10 employees (pre-PMF): Default to speed on nearly everything. Almost all decisions are reversible at this stage. The cost of slow decision-making — missed learning cycles — is catastrophically high. The one exception is co-founder decisions and early team decisions, which have long irreversible tails.
10–50 employees (scaling): Start designing explicit decision authority. The informal communication structures that worked at 10 people break down at 30. People start making assumptions about who owns what. Resolve ambiguity by design, not by conflict.
50–200 employees (mid-stage): The biggest risk is decision centralization in the founder. Push decisions down aggressively. Build judgment in your senior team by letting them make Type 1 decisions that you would have previously owned, with your coaching and review after the fact, not before.
200+ employees: Your primary job as a decision-maker is to design the decision architecture, not to make individual decisions. The quality of your organization's decision system — who decides what, by what process, with what accountability — matters more than the quality of any individual call you make.
Decision-making is a skill, not a trait. It responds to deliberate practice, which means you can get meaningfully better at it if you practice the right things.
The highest-leverage practice areas for founders:
Speed without recklessness: Practice making Type 2 decisions in minutes, not hours. Set a decision deadline. When the timer expires, make the call with the information you have. Review outcomes systematically to calibrate your instincts.
Structural pre-mortem habit: Run a 10-minute solo pre-mortem on every major decision before you bring it to the team. Write down the top three failure modes. This habit alone will catch a significant fraction of the reasoning errors that lead to bad Type 1 decisions.
Explicit dissent invitation: In every team meeting that involves a significant decision, explicitly ask for disagreement before finalizing. "I want the strongest case against this. Who has it?" Make this a habit and people will start coming to meetings having prepared their objections.
Decision review: Block 30 minutes per month to review decisions from 60 and 90 days prior. Make it recurring. Do not skip it when things are busy — that is exactly when the feedback loop is most valuable.
Confidence calibration: For one month, rate your confidence on every decision you make. At the end of the month, check your outcomes against your ratings. Most founders discover they are overconfident in their strong domains and underconfident in their weak ones. Both corrections are valuable.
The best founders I have watched over the years are not the ones who have some innate sense of certainty about their choices. They are the ones who have built a system that produces good decisions consistently — even under high pressure, even under time constraints, even when they are tired and the stakes are high. That system is learnable. The framework in this post is the foundation.
Start with the decision journal. Add the pre-mortem habit for your next big strategic call. Design your decision authority matrix before it becomes a conflict. These three changes alone will move the needle on your organization's decision quality faster than almost any other leadership investment you can make.
How do I know when I am overthinking a decision vs. being appropriately careful?
Ask yourself: "Is the additional time I am taking likely to change my decision, or am I using deliberation to avoid committing?" If more research would change your answer, keep going. If you already know what you are going to decide and are just building a case for it, you are overthinking. The clearest signal is whether you can articulate what specific information would cause you to decide differently. If you cannot, the information gathering is done. Decide.
What do I do when I genuinely don't know if a decision is reversible or not?
Default to treating it as irreversible. The cost of over-deliberating a decision that turns out to be reversible is much lower than the cost of under-deliberating a decision that turns out to be irreversible. When in doubt, slow down and use the pre-mortem. If the pre-mortem surfaces no significant unmitigable failure modes, you likely have more reversibility than you thought.
How do I handle a co-founder who keeps relitigating decisions after they are made?
This is a decision authority and disagree-and-commit problem, not a personal one. First, ensure the decision was made through a process where they had genuine input and that input was documented. If they still relitigate after that, the conversation needs to be about the process, not the specific decision. "We made this decision through a process where your objections were heard and recorded. The appropriate next step is the 60-day review checkpoint, not reopening it now." If the behavior persists, it is a co-founder alignment issue that requires a direct conversation about organizational trust — the Cofounder Conflict Resolution framework provides a structured approach for navigating exactly this kind of escalating disagreement.
Is it ever right to reverse a Type 1 decision?
Yes. The classification is about the default process for entering the decision, not a guarantee that it cannot be undone. If new material information emerges that was not available at decision time, reversing a Type 1 decision can be the right call. What you want to avoid is reversing Type 1 decisions due to discomfort with the commitment, social pressure, or recency bias toward new information that does not actually change the fundamental analysis. The test: would the people who made the original decision, given this new information, have decided differently? If yes, revisit. If the new information is just more of the same signal, hold the line.
How do I make good people decisions when I don't have much data?
You almost never have enough data on people decisions, because people decisions involve predicting future behavior in a context that does not yet exist. The antidote is structured process: work sample tests over credentials, reference calls with people who managed them (not who they listed as references), written scorecards that separate "impressions" from "evidence," and explicit expectations documented before the hire with a 90-day review. The goal is not to eliminate uncertainty — it is to make the uncertainty explicit and build in early correction mechanisms. The most expensive people mistake is not making a bad hire; it is keeping a bad hire too long because the correction feels difficult. Before making a first sales or growth hire specifically, the 12-signal readiness test is a useful pre-decision checklist.
How should I think about decisions that the board has strong opinions on?
Distinguish between decisions that are genuinely in board territory — governance, financing, major strategic pivots — and decisions where the board has opinions but it is not their domain. For the former, their input is legitimate and often valuable. For the latter, you need to be clear — respectfully, with supporting reasoning — about what is an operator decision. Boards that make operational decisions tend to make bad ones because they lack the contextual information. Protecting decision authority at the right level is part of your job as CEO. The appropriate response to board overreach on operational matters is not deference — it is transparent communication about why that decision belongs at the operator level, what your process is, and what the outcome was.
What is the single most impactful change I can make to my decision-making right now?
Start a decision journal. It does not require a process overhaul or a team change. It is a private record you keep yourself. The act of writing down what you are deciding and why — and then reviewing it — will surface patterns in your decision quality that are impossible to see in real-time. Most founders who start it report that within three months, they have identified at least one persistent bias that they had no idea was affecting them. It is the single highest-return investment in self-improvement that I know of for founder performance. Start with your next significant decision. Write it down before you make it. Set a 90-day calendar reminder to review. That is the whole system. The rest follows from the practice.
The framework in this post reflects my experience building and investing in startups over many years. The specific tools — OODA loop, pre-mortem, decision journal, disagree-and-commit — have well-established research foundations. The application to the startup context is the layer I have added from direct observation. None of this is theoretical. These are the tools I wish I had had in my first years as a founder.
The premature growth hire is one of the most expensive early mistakes. Complete framework for deciding when, who, and how to hire your first growth person.
A diagnostic checklist for founders in years 1-3 — 42 checks across product, market, team, revenue, and operations with a scoring framework and recovery playbook.
A founder's guide to recognizing burnout archetypes, diagnosing your current state, and building a sustainable operating system that keeps you in the game.