- What Bezos Actually Said
- Where the Maxim Runs Out
- The Four Upgrades the System Makes
- Why the Difference Matters on Your P&L
- Complement, Not Combat
- Running Both on Monday
- The Book Bridge
- About the Stagnation Assassin
You probably arrived here knowing one thing: Jeff Bezos said something about making decisions with 70 percent of the information you wish you had. You want to know what he meant, whether it holds up, and how to actually use it.
Short answers: he meant that waiting for 90 percent usually means moving too slow, the idea absolutely holds up, and you cannot actually use it as stated, because a maxim is not a system. A sentence in a shareholder letter tells you the destination. It does not hand you the vehicle. This page gives you both: the Bezos version, presented accurately and generously, and then the operator-grade system that turns the maxim into an operating tempo your whole organization can run.
What Bezos Actually Said
Bezos’s 70 percent rule comes from Amazon’s 2016 letter to shareholders: most decisions should probably be made with around 70 percent of the information you wish you had, waiting for 90 percent usually means being slow, and good course correction makes being wrong less costly than you think, while slowness is expensive for sure.
The formulation sits in the letter’s section on high-velocity decision making, one of the defenses Bezos lists for staying a “Day 1” company. Paraphrased faithfully: most decisions should probably be made with somewhere around 70 percent of the information you wish you had. If you wait for 90 percent, in most cases you are probably being slow. Either way, you need to be good at quickly recognizing and correcting bad decisions, because if you course-correct well, being wrong may be less costly than you think, while being slow is expensive for sure.
The same letter carries two companion ideas that made the whole framework famous. First, the two-door distinction, which the letter invokes and which Bezos laid out in more detail in his 2015 letter: some decisions are one-way doors, consequential and nearly irreversible, deserving slow deliberation, while most decisions are two-way doors that can be walked back, and treating two-way doors like one-way doors is how large organizations grind to a halt. Second, disagree and commit: the mechanism for moving forward at speed without requiring consensus.
Give the man full credit. This is the best-known articulation of decision velocity in modern business, it is directionally correct on every point, and it diagnosed the large-company disease, one-size-fits-all heavyweight decision processes, more crisply than a shelf of change-management books. The maxim is right.
The maxim is also incomplete, and the incompleteness shows up the moment you try to run it on a Tuesday.
I go deeper on this in my podcast episode on the 70% Rule and analysis paralysis.
Where the Maxim Runs Out
Bezos’s maxim is right but leaves four operating questions unanswered: how you know you are at 70 percent, which decisions get the 70 percent treatment, how the rule runs below the CEO, and what happens when a 70 percent call goes wrong. Where Bezos gives a maxim, the operator needs a system.
Try to operationalize the shareholder-letter version and four questions appear immediately, and the letter answers none of them.
How do you know when you are at 70 percent?
Feeling 70 percent sure is not the same as having 70 percent of the ideal information, and the gap between those two things is where most bad fast decisions live.
Which decisions get the 70 percent treatment? “Most” is not a sorting mechanism. The two-door metaphor gestures at the answer, but a binary metaphor cannot classify the real decisions on a real desk, most of which are neither purely one-way nor purely two-way.
How does the rule run below you? A founder-CEO can decide at 70 percent by personal fiat. An organization cannot, unless the thresholds, the classifications, and the tolerance for calibrated misses are built into the operating rhythm. A maxim scales to exactly one calendar.
And what happens when a 70 percent decision goes wrong? Because some will, at a knowable rate, and an organization that punishes those misses will quietly revert to 95 percent behavior within two quarters no matter what the CEO quotes at the town hall.
Where Bezos gives a maxim, the operator needs a system. Here is what the system adds.
The Four Upgrades the System Makes
The 70% Rule system makes four upgrades to the maxim: calibration through the Three-Question Test instead of a feeling of confidence, a four-class threshold table instead of a two-door binary, organizational deployment through delegated decision rights and a published cadence, and an explicit error budget that corrects calibrated misses instead of punishing them.
Upgrade one: calibration, not confidence
The 70% Rule defines the threshold against ideal information, the information a decision-maker with unlimited time and budget could gather, not against how sure you feel. And it replaces the feeling with a mechanical gate: the Three-Question Test. Do you understand the two or three material risks that could impair the decision? Can you explain the decision cleanly to an intelligent outsider? Do you have a hypothesis specific enough to be tested against the actual outcome? Three yes answers and the threshold is reached. The test is the threshold, not a checkpoint on the way to a higher one. Keep gathering analysis after passing it and you are working against the rule.
Upgrade two: a threshold table, not a binary
The two-door metaphor becomes the Decision Classification Matrix: four classes built on reversibility and criticality, each with its own calibrated threshold. Reversible and non-critical decisions move at 50 to 60 percent under a forty-eight-hour guarantee. Reversible and critical decisions (the pricing moves, the portfolio changes, the territory restructures) live at the 70 percent threshold that gives the rule its name. Irreversible but non-critical decisions move at 70 percent with exit clauses engineered in. And irreversible, critical decisions (the acquisitions, the plant closures, the senior leadership changes) earn 85 to 90 percent and the patience that requires. One number becomes a calibrated instrument.
Upgrade three: organizational deployment
The system pushes the thresholds down through decision rights, so the organization’s velocity is not capped by the operator’s calendar. Decision classes get delegated with the rule attached, reviews measure velocity without re-litigating individual calls, and the cadence gets published: weekly for the fast class, monthly for the 70 percent class, quarterly for the big ones. This is the difference between a CEO who decides fast and a company that does. Martin G. Moore makes a related point in Harvard Business Review’s How to Make Great Decisions, Quickly: great decisions get made as close to the action as possible, and the most powerful people in a company are rarely the ones doing the hands-on work. If every decision still routes through one office, you do not have a velocity culture. You have a Founder Bottleneck with a good quote on the wall.
Upgrade four: the error budget
A team running at 70 percent will be wrong at a knowable rate. That is not a flaw in the system. That is the system, and the misses are the fee you pay for the tempo.
So the system makes the budget explicit: calibrated misses on reversible decisions get corrected and studied, never punished, because punishing a well-classified, well-tested miss teaches the organization to sandbag its confidence and pad its analysis, and the queue you dissolved quietly reforms. Bezos hinted at this with course-correction. The system makes it a stated policy, because unstated tolerances do not survive the first bad quarter.
Why the Difference Matters on Your P&L
Slow decisions are a structural variable that damages the income statement in order: a queued decision compresses Flow-Through immediately, Profit Velocity next as unmade pricing or mix moves ship at old economics, and Revenue Velocity last when a faster competitor takes the account. Decision velocity is a slope input, not a personality trait.
Here is the part no version of the maxim covers: slow decisions are not a cultural annoyance. A queued decision compresses Flow-Through immediately, because the people who built the request carry overhead while producing nothing. It compresses Profit Velocity next, as the unmade pricing or mix move keeps shipping at the old economics. And it hits Revenue Velocity last and loudest, when the competitor who moved faster takes the account.
I have watched that sequence run in reverse. In a refrigeration price war, we raised prices 8 percent at 70 percent confidence, built on competitor financial distress visible in public information, a customer survey showing price was not the primary driver, and basic margin math showing the current path was unsustainable. Competitors followed within two weeks. The decision saved roughly five months against the traditional analysis cycle and was worth $90 million in reduced losses.
Decision velocity is not a personality trait. It is a slope input, which is why it belongs inside a diagnostic framework and not just a shareholder letter.
Complement, Not Combat
The 70% Rule complements and extends Bezos’s maxim rather than attacking it. The Bezos formulation is the right instinct, famously stated; the 70% Rule is that instinct built into a classification matrix, a mechanical test, a delegation architecture, and an error budget. The maxim starts the conversation. The system survives contact with Monday.
To be completely clear about the relationship: this is complement-and-extend, not attack. The Bezos formulation earned its fame. The 70% Rule is what the instinct looks like after an operator builds it into that architecture and then runs a company on it.
The full framework (the matrix, the Three-Question Test, the Time-Value Calculation, the reversibility discipline, and the cadence architecture) lives in the 70% Rule guide linked above. Start there. Your decision queue is compounding while you read.
Running Both on Monday
Translate the maxim into one working week: Monday, classify every pending decision by reversibility and criticality; Tuesday, run the Three-Question Test on the reversible ones and decide those that pass; Wednesday through Friday, execute and write a dated, testable hypothesis for each call. Then publish the thresholds and state the error budget.
If you walked in with the maxim, here is the translation into a working week, because the fastest way to feel the difference between a quote and a system is to run one.
Monday: list every pending decision on your desk and classify each: reversible or not, critical or not. The two-door instinct you already had from the shareholder letter does the first sort; the reversal-cost math does the second, honest pass.
Tuesday: run the Three-Question Test on everything that landed reversible. The ones that pass get decided this week, and the routine ones get decided in forty-eight hours.
Wednesday through Friday: execute, and write down the hypothesis for each call, one sentence, testable, dated, because the hypothesis is what converts a fast decision into a learnable one.
Then the two moves the maxim never mentions. Publish the thresholds to your team, with the decision classes they now own outright, so the velocity stops being your personal property. And state the error budget in the same breath, out loud, before the first miss arrives: calibrated misses on reversible calls get corrected, not prosecuted. That announcement costs one minute and determines whether the whole system survives its first bad outcome.
Run that week once and you will notice the thing the letter could not give you: the queue clears, the tempo becomes visible to the organization, and the 70 percent number stops being something you quote and starts being something you run.
The Book Bridge
In Ten Minute Transformation, the 70% Rule is the operating tempo of the rebuilt operation: the chapter that installs decision velocity after the structural upgrades and before the strategic navigation work. The book carries the complete system, including the cart company arc where five structural moves were made at roughly 70 percent confidence.
Those moves were made inside the window that made them worth making. Get Ten Minute Transformation and turn the maxim into your operating rhythm.
About the Stagnation Assassin
Todd Hagopian is a Fortune 500 transformation executive who has generated $3B in shareholder value across Berkshire Hathaway, Illinois Tool Works, Whirlpool, and JBT Marel, where he serves as Global VP of Product Strategy and Marketing. Known as The Stagnation Assassin, he is the author of two published books: The Unfair Advantage: Weaponizing the Hypomanic Toolbox and Stagnation Assassin: The Anti-Consultant Manifesto. His blog is published in 15+ languages and read by operators worldwide. Bring him to your stage via his speaking page or connect with him on LinkedIn.
You have the maxim. Find out whether you have the system. Bring your pending-decision list to a free 15-minute Decision Velocity Audit: we sort every item into its class, run the Three-Question Test on the reversible calls, and find out exactly how much certainty your organization is over-buying. Run the Decision Velocity Audit with Todd.