Sizing investment under uncertainty (incl. reversibility / one-way doors).
what we mean by this
Bet is not about being bold. It's not about "taking smart risks" or "moving fast." It's about two specific questions — both required, both routinely conflated — that determine whether you're deploying capital (time, attention, headcount, runway) intelligently or just confidently.
The first question: *how much are we putting in?* The second question: *how easy is it to get out?*
Most PMs answer the first question and stop. They size the bet in terms of engineering weeks or sprint allocation. What they skip is the reversibility read — whether the decision creates structural lock-in, whether it closes off adjacent possibilities, whether the cost of being wrong is bounded or open-ended. A reversible $500K bet and an irreversible $500K bet are not the same bet. They have the same dollar sign but completely different risk profiles.
The Bezos two-way-door / one-way-door framing — which you probably know — is correct as far as it goes. What it doesn't tell you is that most product decisions aren't self-evidently one kind or the other. The doors are often disguised. What looks like a two-way door (we can always roll this back) is frequently a one-way door in practice (you can technically roll it back, but the team dependency, the customer expectation, or the data loss makes rollback prohibitively expensive). Bet, at high levels of acuity, is about seeing through that disguise.
The wrong beginner take on Bet is that it's a sizing judgment — how big should this be? That's necessary but not sufficient. The full competency is: *given the size and given what I know about reversibility, how should I structure this investment to maximize the option value of being right and minimize the cost of being wrong?*
citation
PL Standard v3.1 · Acuity · Bet
Markdown form: [PL Standard v3.1 · Acuity · Bet](https://pragmaticleaders.io/framework/competencies/bet)
the four levels
Anchored at every rung. The blockquote at each level is panel-authored and pulled live from the rubric — edit anchors via the panel tooling and they appear here.
L1Developing
sizes bets by gut or copying past projects; skips reversibility checks. when pressed to justify commitment level, needs a manager to reframe or reduce the ask.
you'll see this when…
A PM allocates 2 weeks to every project in the quarter regardless of what each one is trying to learn or prove.
They describe the bet entirely in implementation terms ("this will take 3 engineers") with no read on what changes if the bet is wrong.
Reversibility concerns are dismissed as hypothetical: "we'd just undo it if it doesn't work" — without examining whether that's true.
common failure mode: The calendar trap — every bet gets the same time box because that's what the sprint structure gives you, not because that's what the uncertainty warrants.
L2Competent
sizes bets on familiar decisions by checking reversibility and adjusting spend, but seeks sign-off when stakes rise or the situation falls outside past experience.
you'll see this when…
A PM explicitly flags which decisions in a planning cycle are two-way versus one-way doors before the team commits.
They structure larger bets with staged checkpoints — "we'll run this for 6 weeks and review before going further."
They can say, plainly, "this one we should move fast on because we can walk it back; that one we need to be more careful because the migration cost is high."
common failure mode: The reversibility read stops at technical reversibility — they catch the code-level rollback risk but miss the organizational, contractual, or customer-expectation lock-in that makes a "technically reversible" decision practically irreversible.
L3Proficient
sizes bets on novel decisions using explicit reversibility checks, names the one-way doors before commitment, and adjusts investment based on what new evidence would change the call.
you'll see this when…
A PM takes 80% of a quarter on one bet because it's clearly reversible — while running four parallel small bets on moves that look irreversible, spending little to learn a lot before committing.
They surface the reversibility disguise explicitly: "this looks like a two-way door but the customer SLA we're signing effectively locks us in — let's treat it as one-way."
They argue, with evidence, for *not* doing a bet cheaply when cheap execution degrades the signal quality: "a half-hearted test here will teach us nothing useful."
common failure mode: The sizing is sharp but the conversation stops at team planning — a proficient PM doesn't always push the irreversibility calculus up to leadership, which means the org keeps making L1 bets at higher altitude even as the PM is correct at team level.
L4Expert
reframes what counts as a good bet — teaches the team to size commitments against reversibility, not confidence, so the org builds fewer expensive traps.
you'll see this when…
They catch someone inflating a $2M bet's strategic frame to get approval — and redirect to the actual reversibility question the approval process was supposed to answer.
They design bets that create option value as a first-class output: "the goal of this quarter isn't just to ship, it's to be positioned to go all-in on the platform bet in Q3 if these four things are true."
They set exit conditions *before* the bet is placed and hold to them even when the bet is running — not because they're rigid, but because they know that exit conditions set after commitment are motivated reasoning, not judgment.
common failure mode: Portfolio paralysis — an expert who's seen too many bets go wrong starts over-modeling the bet landscape, mapping reversibility, sizing implications, option trees, and delays actual bet placement. The competency is calibrated investment under uncertainty, not perfect investment under certainty. The L4 trap is treating the modeling as the work.
how to develop it
The fastest move is to audit your last three quarters for calendar-trap behavior. Print the roadmap. Ask: was each item sized based on what we needed to learn, or based on what the sprint gave us? Were any of the "experiments" actually irreversible in practice? The gap between what you assumed and what was actually true is your development surface.
Read. [Manual: Prioritization and Roadmap](/manual/core-skills/strategy/prioritization) — the sizing half of Bet lives here. [Manual: Product Vision and Strategy](/manual/core-skills/strategy/product-vision-strategy) — how bets compose into a portfolio with coherent option value. [Manual: Experimentation](/manual/core-skills/execution/experimentation) — structuring reversible tests correctly.
Practice. Scenario type: you're given a roadmap where three items have similar expected value but different reversibility profiles — restructure the investment. Scenario type: a decision is being presented as two-way; name why it's actually one-way and propose an alternative structure.
Write. Brief prompt: *You have a quarter to invest. One large bet on a known opportunity. Four smaller bets on uncertain but potentially irreversible moves. Walk us through how you size and sequence them, and what would cause you to abandon the large bet early.*
Coach yourself. Before your next planning cycle, write down — for each item — "what would it take to undo this if we're wrong six months from now?" If the answer is "I'm not sure," that's a one-way door you haven't read yet.
how to spot it in others
They bring reversibility into conversations unprompted — not as a risk disclaimer but as a structuring argument for *how much* to invest.
When asked to scope something, they ask "is this reversible?" before they ask "how long will it take?"
They size experiments differently from commitments without needing you to tell them to.
They can name the exit condition for a bet already in motion — not as something they'd calculate if asked, but as something they're already tracking.
They resist the urge to "just run a small test" on genuinely irreversible decisions — and can explain why half-measures on one-way doors produce worse signal than clarity about direction.
three failure modes we see often
The all-in feint. A PM frames a $3M platform rewrite as a "strategic transformation" to get executive approval, when the actual bet is 80% sunk-cost rationalization and 20% genuine opportunity. The inflated strategic frame papers over the reversibility question — which is: we are about to lock the org into a three-year architectural constraint that we can't unwind without significant regression. Calling it bold doesn't change the door type. This failure mode is partly organizational (approvals reward bold framing) and partly a judgment failure — the PM has stopped distinguishing between confidence in a direction and evidence that the door type warrants the investment.
The reversibility lie. This one is often not intentional. A PM presents a decision as two-way because technically you could reverse it — the feature can be toggled off, the contract has an exit clause, the platform has a migration path. What they don't price in: the customers who've built workflows on the feature, the goodwill cost of the exit clause, the six months of migration that would consume the whole engineering team. Reversibility has to be measured at the real cost of reversal, not the theoretical possibility of it. A door with a ₹2 crore reversal cost is a one-way door, even if the key still works.
The small-bet hoard. Some PMs, especially after being burned by a big bet that didn't pan out, develop a defensive posture — only making small, defensible bets because they're easy to justify and easy to kill. The result is a roadmap that looks busy but never creates the concentration of investment needed to actually move a metric. Small bets are right for irreversible moves and learning missions. But on a clearly reversible opportunity — where the asymmetry strongly favors concentrating — hoarding small bets is a failure of nerve dressed up as rigor.
Signal is about reading conflicting evidence to a decision. Bet takes over after the read: once you've processed what the evidence says, Bet asks how much to put behind it and in what structure. You can have strong Signal and weak Bet — you read the situation accurately, then size the response poorly. They're adjacent because evidence quality directly shapes reversibility assessment, but they test different things.
Reframe is about spotting the wrong question before you solve it. Bet is about correctly sizing the answer to the right question. Sometimes the Reframe move is precisely to surface that the team is treating a one-way door as two-way — which is a wrong framing, not just a wrong bet. But once the question is right, Bet is the sizing judgment, not the framing judgment.
Worth is about which bets are even on the table: deciding what's worth building among everything you could ship. Bet is about *how much* to put behind the decision to build. Worth picks the targets; Bet sizes the investment. A PM who is strong on Worth but weak on Bet will correctly identify the right opportunities and then over- or under-invest in them — creating waste or leaving signal uncollected. Worth and Bet compose; neither replaces the other.
what good looks like in the wild
A mid-stage startup, B2B SaaS, six engineers. Q2 planning. Three opportunities on the table: a new integration with a major enterprise platform (high expected value, irreversible once announced to enterprise customers), a pricing experiment on the self-serve tier (high expected value, clearly reversible — toggle a config), and an exploratory bet on a new vertical that could 3x TAM but has no validation yet.
The PM took 80% of the quarter — five engineers, twelve weeks — on the pricing experiment. Not because it was the highest expected value in absolute terms, but because it was clearly reversible. They could run it, learn from it, and if it degraded self-serve conversion, roll it back within a sprint. The asymmetry was strong: full investment, bounded downside, large upside from learning alone.
On the enterprise integration — the one with genuine one-way-door properties — they did a single-engineer spike for six weeks. Not to build, but to answer the specific technical and commercial questions that would determine whether the full investment was warranted. No announcement, no sales motion, no customer commitment. The spike would give them a yes/no on the irreversible decision before the irreversible decision was actually made.
On the new vertical, they wrote a two-page brief, ran eight customer conversations, and spent nothing else. The bet wasn't ready to size.
At the end of Q2: pricing experiment had generated a clear signal (negative — the price increase hurt more than expected, rolled back in week 10). The spike on the enterprise integration had surfaced a contract complexity that would have made the full bet far more expensive than modeled. The vertical had two strong customer pull signals that made it worth a small probe in Q3.
“
No heroics. No all-in move. No moonshot narrative. Just a clean read on what was reversible, a proportionate concentration of investment in the reversible bet, and disciplined probing on the irreversible ones. That's what Bet looks like at L3: the quarter's shape is defined by reversibility, not by equal time-boxing or by the highest-expected-value item getting all the resource.