Reversibility is a real project asset that decays on a schedule, and almost no organisation keeps an inventory of it, which is why nobody can say which approval was the last one that could still have been undone.
Eighteen months into a large commitment, a director asks the question that changes the temperature of the room. If we decided today that this was wrong, what would it cost to stop? A number arrives within the week: large, carefully built, defensible. Then the same director asks a second question, and no number arrives at all: when was the last moment we could have stopped for materially less?
Nobody in that room has been negligent. The approval papers are complete. Every variation has passed through change control and been priced. The steering committee has met on schedule and its minutes record decisions rather than discussion. What is missing is not diligence but a particular record: any statement, made at the time of each approval, of what that approval was about to make impossible.
This matters because the enterprise has been spending something valuable without metering it. The capacity to stop, unwind, re-scope or switch suppliers is not a management mood. It is a finite, depleting resource converted into progress, usually without anyone noticing the exchange and almost never with a signature attached to it. By the time the balance is asked for, it has been drawn down by dozens of small, reasonable decisions taken for other reasons entirely.
Most enterprises can produce a cost-to-complete within days and a cost-to-stop within a fortnight. Almost none can produce the date on which stopping ceased to be an option.
The Strategic Context
Project management has taught the underlying dynamic for decades, in almost every introductory treatment. The ability to influence an outcome is highest at the beginning and falls steadily; the cost of changing anything rises as commitments lock in. The teaching is consistent and correct.
What is striking is the conclusion drawn from it. The declining curve is used, essentially without exception, as an argument for front-loading effort: plan harder, control the design phase, put the analytical weight where it buys the most. That is sound advice, and not the only conclusion available. If a quantity begins high, declines predictably and cannot be replenished, the discipline has described an asset with a depletion schedule. Assets of that kind get balances, owners and reporting lines. This one gets neither a register nor a custodian.
The gap is visible inside the teaching material itself. The same treatments that plot the decline describe a pre-implementation review as the last opportunity to step off before serious resources are committed, and insist that every phase boundary carries a real choice to proceed, modify or terminate. Both presuppose a quantity of remaining reversibility; neither comes with any instrument recording how much is left. A discipline that can price a change precisely has no line item for whether the change remains possible.
At enterprise level the omission compounds. A single project consumes its own reversibility. A portfolio consumes it in parallel, across commitments never compared with one another, and the aggregate position, how much of the enterprise's commitment base is now fixed, appears on no report a board receives.
What Leaders Commonly Misread
The first misreading is to treat contractual optionality as reversibility. A termination-for-convenience clause is a legal right. Whether it can actually be exercised depends on what the money has already been turned into, who has been told what, and which physical facts now exist in the world. A right nobody has costed and nobody has exercised is a claim on an asset, not the asset.
The second is to confuse the cost of change with the possibility of change. Change control answers the first question well: every variation assessed, priced and routed regardless of size. But a system that prices changes returns a very large number for an impossible change just as readily as for a merely expensive one. Nothing in it separates a reversal that would be painful from one that has ceased to exist. A change register is a price list, not a stocktake.
The third is to assume the decline is smooth. The textbook curve is continuous; reality is not. Reversibility falls in steps attached to particular signatures: the mobilisation order, the long-lead purchase, the demolition, the public announcement of a date. Between those steps the balance barely moves. At them it can reach zero in an afternoon. The smooth model produces the characteristic executive error of assuming there is still room because the process feels gradual.
Reframing the Issue
The useful move is to stop drawing a curve and start keeping a balance. Reversibility has an opening position at authorisation, consumption events, and a closing position at any date one names. It is spent, not lost. Like every other asset the enterprise holds, it should have a custodian who can be asked for the number.
Consider an offshore aquaculture licensing programme as a hypothetical. The enterprise holds a lease and a staged consent regime. Early on almost everything is reversible: the site can be surrendered, the engineering re-specified, the species reconsidered. Mooring installation is expensive to reverse but genuinely reversible. Then biological stock enters the water, and an entire class of reversal is no longer expensive; it is unavailable. No expenditure returns the site to its pre-stocking state within the relevant timeframe, and no commercial instrument moves that fact to a counterparty. The decision that closed the option was not the largest in the programme, and in most governance packs it would not be the most prominent item on the page.
That is the shape of the problem. The decision consuming the most reversibility is rarely the one consuming the most money, and it is almost never put to a board as a decision about reversibility.
Where the Balance Actually Goes
It is spent by decisions taken for other reasons
The commitments that draw down reversibility fastest are usually taken to buy schedule or price certainty. A long-lead order placed early protects against a delivery window. A site handover accepted ahead of time keeps a contractor's crew engaged. A date announced publicly secures political air cover. Each is a defensible trade in which reversibility is the currency paid, and each is invisible because only one side appears in the paper: the schedule benefit is quantified, the reversibility spent is not recorded.
It is held in three currencies, and they deplete separately
Physical reversibility is the capacity to return the world to a prior state: undemolished, unstocked, un-migrated. Contractual reversibility is the capacity to exit an agreement at a knowable price. Political reversibility is the capacity to change course without a cost in credibility, licence or standing the enterprise cannot bear.
These deplete on different schedules, and the effective position is the lowest of the three, not the average. An enterprise records digitisation programme makes the point as a hypothetical: the vendor contract may stay terminable at modest cost throughout, keeping the contractual balance healthy, while the physical balance reaches zero on the day the source paper is destroyed. Monitoring only the contract means believing in an option the enterprise no longer holds.
Political reversibility is the one most often mistaken for a soft factor. Where a commitment has been made in a public consent process or announced to a community, the practical ability to change course can close well before either of the other paths, and it closes quietly.
It is consumed on a schedule set by the order of work
The sequence in which activities are arranged determines when each balance falls, and most sequencing rules push the hardest and least recoverable work later rather than earlier. That ordering mechanism, how criteria for sequencing systematically schedule irreversibility into the window where the balance is already thin, is the subject of [Related article: From Sequencing Rule to Point of No Return], and this article does not take it up. What concerns us here is the balance itself: its size, its custodian, and the fact that no one is reading it at the moment of signature, whatever order the work happens to be in.
Decision Framework: The Reversibility Inventory
The instrument is a register, not an analysis. It sits alongside the risk register, is refreshed at every gate, and fits on one page. Each entry covers one live commitment and answers five questions.
One. What is the commitment, and what would reversal actually consist of? Not "terminate the contract" but the physical, contractual and political steps that would have to occur, in order, to put the enterprise materially back where it was.
Two. What is the position in each currency today? Use three states only; finer gradation invites debate rather than decision.
| Currency | Available | Costly | Closed |
|---|---|---|---|
| Physical | Prior state restorable within the decision horizon | Restorable at material cost or delay | No expenditure restores it in the relevant timeframe |
| Contractual | Exit priced and tested in the last twelve months | Exit right exists, price not calculated | No exit, or exit priced above the value at stake |
| Political | Course change requires no external explanation | Explanation required to a known party | Change would cost licence, consent or standing |
Three. What is the closing date for each currency: the date after which the entry moves to Closed? A forecast is acceptable and far better than none. An entry with no date is the finding, not a gap in the paperwork.
Four. Who owns the entry? A named individual with authority to declare a currency Closed. Where no such name exists, reversibility is being closed by people who do not know they are doing it.
Five. What was received in exchange? One line stating the schedule, price or political benefit bought by the drawdown.
Two governance tests make the register bind. At every approval the standing question is: which reversal paths does this decision close, and does the paper say so on its face? An approval that closes a currency without stating it is returned unsigned. Retrospectively, take the last three approvals of any size and ask each owner for the closing dates. The speed of the answer is the diagnostic; the answers are secondary.
From Strategy to Execution
Immediate. Take the two or three largest live commitments and reconstruct their closing dates from the existing record. It is a week of work, and uncomfortable: the reconstruction usually shows the balance reaching zero months before the last gate described as a decision. Add one mandatory line to the approval template: what this decision makes impossible.
Medium-term. Assign custodians. Price termination and exit provisions at the point of contract drafting rather than at the point of crisis, and re-price them annually. Where supplier selection is the commitment in question, recognise that the reversibility of that choice may have been exhausted long before the tender opened, by a qualification process that narrowed the field years earlier; how that narrowing happens is the subject of [Related article: The Shortlist Was Decided Years Before the Tender], and is outside the scope of the inventory described here, which measures the balance rather than explaining how it came to be pre-spent.
Long-term. Aggregate at portfolio level, and build the capability to buy reversibility deliberately (a staged commitment, a smaller first tranche, a deferred order accepted at a premium), recording the premium as the purchase of an asset rather than an inefficiency to squeeze out at the next cost review. Often the permission to stage a commitment already exists in the enterprise's mandate and has simply never been exercised; what a mandate already permits, and why executives read past it, is the subject of [Related article: What Are You Already Permitted to Do?], and this article assumes only that the balance should be read, not that a new authority is needed to read it.
Signals to Monitor
The phrase "we are too far in" entering executive discussion without a date attached is the clearest early signal: it is a statement about the balance by someone who has not consulted one. Watch also for approval papers carrying a benefits case and a risk section but no reversal paragraph; for material commitments authorised between gates rather than at them, where most quiet drawdown occurs; and for change registers full of priced variations with no field separating feasible from infeasible.
Two further signals sit at the edges of the record: a termination clause nobody in the enterprise has ever costed, and a closure report that measures performance against scope, budget, schedule and quality while saying nothing about when the option to stop expired. Closure is the last chance to convert that experience into something the next programme can use.
Questions for the Leadership Team
- For our three largest live commitments, on what date did the physical reversal path close, and whose signature closed it?
- Which of our current contracts contains a termination price that has been calculated rather than assumed, and when was each last recalculated?
- Of the last four approval papers put to this board, how many stated what the decision made impossible?
- Who holds the authority to declare a reversal path closed, and if the role does not exist, who has been closing them?
- What did we pay in premium or delay for the last option we deliberately kept open, and was it recorded anywhere as a purchase?
- On the last programme we completed, at what point would a reconstructed balance show zero, and how many gate reviews described as decisions occurred after it?
Closing Perspective
Reversibility will be spent whatever the enterprise does, and spending it is often correct: an organisation that preserves every option indefinitely never commits to anything, and pays for that in a different currency. The choice available to a leadership team is not whether to consume the asset but whether to consume it knowingly, at a moment someone has named, in exchange for something written down.
The responsibility that follows is narrower than it first appears and harder to delegate. It is not to keep options open. It is to ensure that when the last one closes, someone in the enterprise knows it has happened, knows what was received in return, and can say so at the meeting where somebody finally asks.
About EraNorth Insights
EraNorth Insights publishes practical analysis on strategy, projects, operations, transformation and decision intelligence for professional and organisational use. About EraNorth.