Section 1
The five challenges at a glance
Five failure modes surround reversibility in practice. First, one-size-fits-all governance: every decision, from a logo tweak to an acquisition, runs through the same deliberation ritual, which Bezos identified as the signature pathology of growing organizations (Amazon, 2015 shareholder letter). Second, misclassification upward: treating reversible calls as irreversible, usually because the founder confuses emotional weight with structural irreversibility; a pricing test feels permanent but is a two-way door. Third, misclassification downward, the rarer but costlier error: treating genuinely irreversible calls, a partner equity grant, a brand repositioning with a public commitment, a key-client termination, as casual. Fourth, threshold drift: even correctly classified Type 2 decisions stall because no one defined how much information is enough, so the 70% point passes unnoticed while analysis continues (Amazon, 2016 shareholder letter). Fifth, missing commitment norms: decisions get made but not closed, and relitigating consumes the time the fast process saved, the waste pattern McKinsey quantified at 61% of decision-making time used ineffectively (McKinsey, 2019). The table maps each challenge; the three analyzed in depth below are one-size-fits-all governance, misclassification in both directions, and the absence of threshold and commitment discipline, because correcting those three captures most of the framework's value.
Section 2
Challenge one: one-size-fits-all governance is a tax on every decision
Bezos's 2015 letter is precise about the failure sequence: as organizations get larger, there is a tendency to use the heavyweight Type 1 process on most decisions, including many Type 2 decisions, and the end result is slowness, unthoughtful risk aversion, failure to experiment sufficiently, and diminished invention (Amazon, 2015 shareholder letter). The research context makes this more than an aphorism. Bain's decade of work across more than 1,000 companies found decision effectiveness, the combination of quality, speed, yield, and effort, correlates with financial performance, and that excessive effort per decision is itself a drag even when the decision lands correctly (Bain, 2010). The service-firm version is recognizable: a six-person agency where a proposal template change waits three weeks for a partner meeting, where a $400 software purchase needs the same sign-off as a $40,000 hire, where the founder's calendar is the firm's decision queue. Each individual delay seems prudent; the aggregate is a firm whose cycle time on everything is set by its most cautious ritual. The diagnostic question is structural, not cultural: does your firm have any officially different process for decisions of different reversibility? If every decision travels the same path, you are paying Type 1 costs on Type 2 calls, and the bill compounds, because the experiments you never ran are invisible on every report.
Section 3
Challenge two: misclassification cuts both ways
The framework only works if classification is honest, and founders err in both directions. Upward misclassification is the common case: decisions that feel weighty get treated as irreversible when they are structurally reversible. Raising prices on new proposals is a two-way door; you can revert next proposal. Niching the marketing message, restructuring the weekly meeting, testing a retainer offer, all reversible within weeks at modest cost. The tell for upward misclassification is that the imagined catastrophe involves embarrassment rather than locked-in loss. Downward misclassification is rarer and more damaging: equity grants, personal guarantees on leases, public brand repositioning, firing the client who funds payroll, naming a managing partner. These doors close behind you, and Bezos's prescription for them is the opposite of velocity: methodical, careful, slow, with great deliberation and consultation (Amazon, 2015 shareholder letter). The judgment literature adds a warning here: classification itself is a judgment subject to noise, and the same founder will classify the same decision differently on different days (Kahneman, Sibony and Sunstein, 2021). The corrective is a written reversibility test applied identically every time: What would reversal cost in money, time, and trust? Who could reverse it, and within what window? If reversal is possible within a quarter at a cost the firm can absorb, it is Type 2, regardless of how it feels.
Section 4
Challenge three: thresholds and commitment, the two disciplines that make speed safe
Classifying a decision as Type 2 only creates the permission to move fast; two further disciplines convert permission into velocity. The first is the information threshold. Bezos's 2016 letter argues most decisions should be made with around 70% of the information you wish you had, and that waiting for 90% is usually just slowness, because being wrong on a reversible call is cheaper than being slow on all of them (Amazon, 2016 shareholder letter). Eisenhardt's field research supports the underlying logic: her fast deciders did not gather less information, they gathered different information, real-time operational data available immediately, rather than commissioning forward-looking studies that took months and aged badly (Eisenhardt, 1989). The second discipline is commitment. The 2016 letter's companion tool is 'disagree and commit': once the owner decides, dissenters state their disagreement and then commit fully, rather than relitigating or slow-walking. Without this norm, fast decisions are an illusion; the meeting ends but the decision stays open, and McKinsey's finding that most decision time is wasted is substantially a relitigating tax (McKinsey, 2019). Together the two disciplines define a contract: the firm grants Type 2 owners the right to decide at 70%, and everyone, including the founder, grants the decision closure. Velocity without thresholds is haste; velocity without commitment is churn.
Section 5
Innovative solutions
Firms operationalizing the framework well use five mechanisms. One: the reversibility gate. Every decision entering the leadership queue gets classified within 24 hours using the written test, reversal cost, reversal window, reversal authority, and the classification is logged, which suppresses the day-to-day noise in judgment that would otherwise make classification arbitrary (Kahneman, Sibony and Sunstein, 2021). Two: asymmetric tracks. Type 2 decisions route to a single named owner with a deadline in days and an explicit 70% threshold; Type 1 decisions get the full treatment, written one-page case, consultation with at least two outside perspectives, a deliberate waiting period, and founder sign-off (Amazon, 2015 and 2016 shareholder letters). Three: delegation by reversibility, not by department. Authority maps to door type: anyone leading a function can walk through two-way doors in their domain up to a stated spend level, which is how a small firm honors Bain's finding that decision effectiveness depends on putting decisions at the right level (Bain, 2010). Four: the reversal budget. The firm pre-authorizes a quarterly allowance for unwinding Type 2 calls that did not work, making reversals routine operations rather than admissions of failure, which keeps classification honest. Five: disagree-and-commit as an explicit meeting move, invoked by name, logged with the decision, and binding on everyone, founder included.
Section 6
Solution framework
The Two-Door Governance Model installs in four layers. Layer one: classification. The written reversibility test becomes a one-minute ritual at decision intake: reversal cost in money, time, and trust; reversal window; reversal authority. Reversible within a quarter at absorbable cost equals Type 2. Layer two: routing. Type 2 decisions go to the closest competent owner with a deadline measured in days and the 70% threshold stated in writing; Type 1 decisions go to a deliberate track with mandatory consultation and a cooling-off period before signature (Amazon, 2015 shareholder letter). The founder's calendar carries only Type 1 calls and Type 2 calls the founder personally owns, which for most firms shrinks the founder's decision load by more than half. Layer three: closure. Every decision is logged with owner, rationale, threshold used, and a review date; disagree-and-commit is recorded by name; reopening requires materially new information, which protects the time savings from the relitigating tax McKinsey documented (McKinsey, 2019). Layer four: learning. At each review date, the outcome is compared with the expectation written at decision time, and classification accuracy is audited quarterly: which Type 2 calls turned out harder to reverse than scored, and which Type 1 deliberations were theater. The audit feeds back into the reversibility test, so the firm's classification instrument improves with every cycle.
Section 7
Evidence-based action plan
Week one: classify your backlog. List every open and recently deferred decision, apply the written reversibility test to each, and tag them Type 1 or Type 2. Expect the distribution Bezos implies: the large majority are two-way doors currently waiting in a one-way-door queue (Amazon, 2015 shareholder letter). Week two: route and delegate. Assign every Type 2 decision a named owner below the founder where possible, a deadline within ten business days, and a one-sentence 70% threshold defining enough information (Amazon, 2016 shareholder letter). Put the genuine Type 1 items, there are usually two or three, on the deliberate track with scheduled consultation. Week three: install closure. Open the decision log, record every call as it closes, and introduce disagree-and-commit explicitly in a leadership meeting, with the founder modeling it first on a decision where the founder is the dissenter; the norm only takes if the most powerful person submits to it. Week four: set the reversal budget and the audit. Pre-authorize the quarter's unwind allowance, schedule the classification audit for day 90, and start tracking two metrics: median days-to-close for Type 2 decisions, and founder decisions per week. Success at 90 days looks specific: Type 2 cycle time under a week, founder decision load halved, at least one cheap reversal executed without drama, and a leadership team that has stopped paying irreversible-decision prices for reversible-decision risks. For adjacent evidence in this pillar, see [The Uncertainty Advantage: What the Research Says About Small-Firm Agility vs Large-Firm Resources](/blog/growth-uncertainty-advantage-small-firm-agility) and [The Weekly Operating Rhythm: What the Evidence Says About Meeting Cadence, Decision Rights, and Execution](/blog/growth-weekly-operating-rhythm-meeting-cadence-decision-rights).