The Decision Journal: Why Top Founders Track Every Big Call
The decision journal is the single highest-leverage habit a founder can adopt, and almost nobody does it. Here is exactly how to keep one, why it works, and the compounding value over a five-year arc.
The decision journal is the highest-leverage habit a founder can adopt. The investment is two minutes per decision. The compounding value over five years is enormous. Almost nobody does it. The disconnect between cost and value is one of the strangest puzzles in operating practice.
Here's exactly how to keep one, why it works, and the specific reasons most founders skip it.
What a Decision Journal Actually Is
A decision journal is a written record of the consequential decisions you make, the reasoning at the time, and the actual outcome. That's the entire structure. It is not a project management tool, a status report, or a strategy document. It is a personal log of your decisions as they were made, kept honestly, reviewed periodically.
The format that works:
At decision time, write four things (under two minutes): 1. The decision (one sentence) 2. The reasoning behind it (three sentences) 3. The key assumptions you're betting on (two or three bullet points) 4. What you'd need to see to know you were wrong (one sentence)
At outcome time (3-12 months later), write three things (under one minute): 1. What actually happened 2. Were the original assumptions right or wrong? 3. What pattern can you extract for next time?
That's the whole system. The simplicity is the point. Anything more elaborate fails to survive contact with a busy founder.
Why This Works
Three structural reasons.
It separates decision quality from outcome quality. A good decision can produce a bad outcome (the risks you accepted materialized). A bad decision can produce a good outcome (you got lucky). Without a written record of your reasoning at decision time, you cannot tell the difference. You'll attribute good outcomes to good decisions and bad outcomes to bad luck — which is the exact wrong calibration.
Annie Duke's Thinking in Bets makes this case extensively. The poker player learns to separate decision quality from outcome quality by analyzing the decision before the outcome is known. Founders need the same discipline, and a decision journal is the cheapest way to install it.
It forces explicitness about assumptions. Writing down "what I'd need to see to know I was wrong" forces you to articulate the falsification criteria for your decision. This is hard. Most founders never do it. The act of writing the sentence sharpens the decision itself — many founders will look at the sentence they're about to write, realize they have no good answer, and recognize they don't yet understand the decision well enough to make it.
It creates a learning loop. Without a journal, you make the same decision errors over and over because you don't have data on them. With a journal, you can spot patterns: "I systematically underestimate transition cost on pivots," or "I systematically overestimate engineering team velocity on new features," or "I systematically commit to hires too fast when I'm tired."
Why Most Founders Skip It
The puzzle is that the cost is two minutes and the value is enormous, and yet decision journals are vanishingly rare in operating practice. Three reasons.
It feels redundant in the moment. When you make a decision, you have just spent hours thinking about it. Writing down the reasoning feels like duplicate work. It isn't. The reasoning that's in your head at decision time will be irrecoverably gone in six months. The write-down is what preserves it.
The payoff is delayed. The compounding value of a journal arrives over years. The cost arrives now. Most founders prioritize what's urgent over what compounds.
It requires admitting uncertainty in writing. Writing "the key assumptions I'm betting on are..." is psychologically expensive because it acknowledges that you're making bets, not certainties. Many founders prefer the illusion of certainty to the written record of their bets.
What the Compounding Looks Like
Suppose you start a decision journal today and keep it religiously for five years. Conservative estimate: 80 decisions logged per year × 5 years = 400 decisions. Of those, perhaps 30 are major strategic calls.
Over five years, you'll have a personal dataset of: - Your personal base rates on common decisions (pivots, hires, pricing changes, fundraises) - The specific cognitive biases you exhibit (which categories of decision you over- and under-weight) - The recurring assumptions you keep getting wrong - The patterns of decision-making that consistently produced good outcomes
This dataset is more valuable than any external advisor's pattern recognition because it's specific to you. A consultant has pattern data on companies generally. A decision journal has pattern data on you specifically.
After five years of journal-keeping, you're roughly 20-30% more accurate on the categories of decision you've journaled most. That's not a guess — it matches the calibration improvement research on expert forecasters who keep written records of their predictions.
How to Start This Week
The minimum viable decision journal takes 10 minutes to set up and works indefinitely.
Pick a tool. Notion table, Airtable, even a Google Doc. The format matters less than the consistency. Avoid anything that requires more than 30 seconds to capture a new entry.
Define your triage rule. What level of decision goes in the journal? Suggestion: anything you'd discuss with a co-founder or advisor before deciding. Anything below that line, don't log — the friction will kill the habit.
Make the first entry today. Pick the most consequential decision you made this month, log it retrospectively. Just to break the activation barrier. (Note: retroactive entries are useful for triage practice but don't count for the calibration loop — your reasoning at the time is what matters, and you won't remember it accurately a month later.)
Set a review cadence. Quarterly review of the last three months' decisions. Note which assumptions proved right, which proved wrong, what patterns are emerging. The review is what turns the journal from documentation into a learning loop.
A Specific Example
A founder logs the decision to hire their first sales rep in March:
Decision: Hire a SaaS sales rep at $120K base + $80K OTE, starting May 1. Reasoning: We have 18 customers via founder-led sales; sales cycles are 6 weeks; founder is the bottleneck; the next 20 customers should be repeatable; first hire de-risks the path to 50 customers. Assumptions: (1) The sales process is repeatable enough to teach a non-founder. (2) A sales rep at this comp level will source 4-6 deals per month within 90 days. (3) We have enough pipeline coverage for a new rep to learn on. Falsification: If the new rep is at 0-1 closed deals at 90 days, the sales process is not yet repeatable and I'm scaling too early.
Six months later, the founder reviews:
Outcome: Rep closed 2 deals in 90 days, 4 in 120 days, 7 in 150 days. Below plan but trending right. Assumption review: (1) Repeatable: partially. The rep needed me to close the last call on every deal for the first 60 days. (2) 4-6 deals: missed. Actual ramp was 2 months slower than projected. (3) Pipeline: yes, this was correct. Pattern: I consistently underestimate ramp time for first sales hires. Next sales hire, plan for 5 months to full productivity, not 3.
That pattern — "I consistently underestimate ramp time for first sales hires" — is worth six figures the next time this founder hires a sales rep. It cost two minutes to capture.
How SynthBoard Helps
For high-stakes decisions, run a structured boardroom session before deciding. The session output makes the journal entry easier — the assumptions and falsification criteria come pre-articulated from the analysis. Pair the journal with the session and you have the full decision intelligence loop: structured analysis at decision time, written record of reasoning, periodic review of outcomes, learning loop closed.
Common Mistakes
Logging too much. If you log every decision, you'll abandon the journal within three weeks. Triage ruthlessly to consequential decisions only.
Skipping the falsification sentence. "What would I need to see to know I was wrong?" is the highest-value question in the journal. Without it, you can't tell whether your assumptions held — you'll just rationalize whatever outcome happened.
Never reviewing. The journal without the review is documentation. The review is what makes it a learning system.
Defending decisions in the review. The point of the review is honest calibration. If you're explaining why each decision was actually right, you're not learning — you're rationalizing.