Most strategic commitments inside large organizations are sized by political gravity, not by the underlying risk shape. The size of the bet correlates to the rank of the executive sponsoring it, the prior year's allocation, and how much capital is left in the budget. None of those are signals about how much of the firm's future the commitment should consume.
The Stake Sizing Matrix is a discipline for fixing that. It is deliberately simple — a two-by-two — because the value is in the forced conversation, not in the analytics. The two axes are reversibility and conviction. The cells are four standard sizes of commitment. Used consistently across a leadership team, the matrix becomes the shared vocabulary for sizing every meaningful bet the company makes.
The two axes
Reversibility
Reversibility is the question of how cheaply the commitment can be unwound if the scenario underneath it does not converge. A new product line built on existing capacity is highly reversible. A long-lived asset purchase, a multi-year supply contract, a brand repositioning, or a workforce restructure is not. The right question to ask is "if we are wrong about this in eighteen months, what does it cost to step back."
Reversibility is not the same as cost. A small but irreversible commitment — a public position-taking, for example — may be cheaper in capital than a large but reversible one. The matrix treats those as fundamentally different shapes of bet.
Conviction
Conviction is the team's calibrated confidence that the scenario underneath the commitment is the one converging. Not "we believe in this." Calibrated. If the team has been wrong before on scenarios that felt this confident, the conviction reading is lower than it feels.
Conviction is read at the team level, not the individual level. Individual conviction concentrates around the loudest voice. Team-level conviction, structured through a brief blind-vote protocol before discussion, is one of the cheapest discipline upgrades any leadership team can install.
The four cells
The matrix produces four standard commitment shapes. Each has a different role in the portfolio.
Probe — low conviction, high reversibility
A small, deliberately reversible bet placed to generate signal. The point of a probe is not return, it is information. Probes should be priced as research expense, not as initiative. Common forms: a regional pilot, a single-customer trial, a posted position to test market response.
Hedge — low conviction, low reversibility
A commitment of meaningful magnitude made when the underlying scenario is uncertain but the cost of being unprepared is structural. Hedges are deliberately paired — one against each plausible scenario — and sized to minimize the regret of any single scenario converging. Common forms: dual-source manufacturing, parallel technology platforms, redundant talent pipelines.
Build — high conviction, high reversibility
A material commitment made when the team has high calibrated conviction in the scenario but the bet itself can be unwound at modest cost. Build commitments fund execution at scale without betting the firm. Common forms: capacity expansion in established markets, product-line extensions, geographic expansion to adjacent territories.
Bet — high conviction, low reversibility
A defining commitment. The firm's future bends around it. Bets should be rare, deliberate, and supported by both high team-level conviction and a public theory of the scenario underneath them. A firm that makes more than one or two Bets in a planning cycle is not making Bets — it is making poorly disciplined commitments with a confident label.
Using the matrix in a Stake decision
At the end of a Predictive Planning Loop revolution, the team arrives at the Stake phase with a set of scenarios and a set of candidate commitments. The matrix is run for each candidate.
First, the team places the commitment on the reversibility axis. This is a structural read — it should not vary by who is in the room. Second, the team places the commitment on the conviction axis using the blind-vote protocol. The variance in individual conviction reads is itself a signal — if it is wide, the team is not yet ready to commit at high size.
The matrix then names the commitment shape. If the team is calling the commitment a Bet but the matrix places it as a Hedge, the conversation is no longer about the commitment — it is about which of the team's reads is wrong. That conversation is the entire point of the matrix.
Worked example
A regional industrial distributor faces a question about electrification. The team has run the loop and arrived at three scenarios: rapid electrification of the customer base in five years, slow electrification over fifteen years, and a fragmented path with electrification leading in two of four end markets.
Three commitments are on the table. Acquire an electrical distributor. Build an internal electrical capability. Stand up a partnership with an electrical specialist.
The acquisition is low reversibility — once paid for, it is on the balance sheet. Conviction is mixed — the team is split on which scenario is converging. The matrix places it as a Hedge or a Bet depending on conviction. If the team votes blind and conviction comes back high and tight, it is a Bet. If conviction is split, the matrix says the firm is not ready to make this commitment as an acquisition — it should restructure the bet to something more reversible while it improves its read.
The internal build is high reversibility — it can be slowed, re-pointed, or shut down with modest cost. Conviction can be moderate. The matrix places it as a Build. The firm can fund it and learn.
The partnership is high reversibility — most well-structured partnerships can be unwound. Conviction is low. The matrix places it as a Probe. The firm should set it up, fund it small, and treat it as a signal generator.
The decision the team arrives at is a portfolio: a Probe and a Build now, with the option to convert to a Bet on the acquisition once conviction tightens. That portfolio shape is what predictive planning produces — and what no amount of traditional capital planning produces by default.
Why the matrix matters
Most leadership teams have no shared vocabulary for the size and shape of a commitment. "Strategic initiative" covers everything from a regional pilot to a generational bet. The matrix forces a team to be specific about what it is doing.
The discipline is not analytical. It is conversational. Two executives who disagree about whether a commitment is a Hedge or a Bet are not having a sizing argument — they are having a conviction argument, and the matrix surfaces that. A firm that regularly mistakes Bets for Builds is taking on more risk than it thinks. A firm that regularly mistakes Probes for Hedges is spending information dollars and getting commitment dollars in return. Either pattern is expensive. Both are correctable.
The matrix is one page. The conversation it forces is what makes a team predictive instead of merely active.