The right to change your mind (flexibility, commitment and what each one costs)

The perfect park

Euro Disney opened in April 1992 with the confidence of a company that knew how to build parks: Disneyland (1955), Walt Disney World (1971) and Tokyo Disneyland (1983) had been successes of increasing scale. The European plan was built on that experience: visitor projections, hotel stays, per-head spending on food and merchandise —all of it calibrated on the behavior of the earlier parks. Europeans did something else: they came (attendance didn't miss by much), but they came differently —day trips instead of long stays, fewer hotel nights, less spending inside, a price sensitivity that Orlando's visitors didn't have. Every small deviation multiplied by millions of visitors: by 1994 the project was renegotiating its debt, and the case had entered the textbooks.

Rita McGrath and Ian MacMillan used it as the opening piece for an idea this article develops: the Euro Disney plan was not a stupid plan —it was a conventional plan applied to an unconventional project. Traditional planning works when accumulated experience is representative of the new ground: it projects from the known, fixes the number, and measures success by compliance. On new ground, that method turns assumptions into facts through the mere act of writing them into a spreadsheet. The ratio between what is known and what is assumed is the variable that decides which kind of plan applies —and nobody calculates it.

This connects with the close of Part 1: if individual advantages last less and less, then every strategy is, more and more often, an incursion into ground where your own experience isn't enough. The question this second part opens is the one that leaves hanging: how much to commit, how much to keep open, and with what tools to decide?

The tension: commitment creates value (and kills)

It is worth starting by defending the accused. Intellectual fashion since the 1990s says "flexibility"; but Pankaj Ghemawat showed in a classic book (Commitment, 1991) that the heart of strategy is exactly the opposite: valuable positions are built out of irreversible commitments —the plant nobody else dares to build, the brand that took twenty years, the distribution network, the accumulated learning. If everything you do is reversible, anyone can copy it; what cannot be imitated is born, almost always, of having burned the ships. The scale advantage of a competitor who committed capacity first is not neutralized with agility: it is suffered.

The problem is the other side of the same blade. The commitment that creates the advantage is the one that later makes it impossible to let go. Dorothy Leonard-Barton gave it its definitive name: a company's core capabilities —its mastered technique, its systems, its engineering values— become, when the environment turns, core rigidities: the very excellence that made it win is what keeps it from seeing and doing the new thing. It is no accident that this happens to the best: it happens because they are the best at the previous game. And the previous series added the psychological layer: overconfidence makes committing feel cheaper than it is (the 0.5% forecasts from Part 1 of that series), and loss aversion plus escalation make letting go feel more expensive than it is (Staw's US$13.1 million). Bias pushes simultaneously toward over-committing and under-decommitting.

The answer to this tension is not a lukewarm middle ground. It is a change of unit: to stop arguing about "are we flexible or committed?" —a question of identity, with no answer— and start managing a portfolio in which some things are committed all the way and others are deliberately kept open. The volume edited by Gary Hamel, C. K. Prahalad and colleagues in 1999 (Strategic Flexibility: Managing in a Turbulent Environment) framed the era: in environments of creative destruction, the ability to renew competences is worth more than any single position. But the concrete tool for managing the portfolio came out of finance.

Real options: buying the right to decide later

A financial option is the right —not the obligation— to buy something tomorrow at a price fixed today. You pay a premium to keep the door open. Timothy Luehrman proposed looking at strategy that way: not as a sequence of projects to approve or reject, but as a portfolio of real options —small investments that buy the right to invest seriously later, once part of the uncertainty has resolved. The pilot in a new market, the minority stake in the emerging technology, the two-person team exploring the channel: these are not "projects with a poor ROI"; they are option premiums whose value lies, precisely, in what they allow you to decide further down the road.

Options reasoning changes three managerial reflexes:

It changes what gets approved. Instead of approving the complete plan (and turning it into an irreversible commitment on day one), you approve the stage that buys the most valuable information. The committee's question stops being "does the NPV of the whole project work?" and becomes "what do we learn with the next dollar, and what is knowing that worth?". It is the logic the previous series applied to complex terrain —small, reversible, staged bets— with its financial foundation made explicit.

It changes how uncertainty is read. For the conventional plan, uncertainty is the enemy: it shrinks expected value. For an option it is fuel: the more uncertain the ground, the more the right to wait and decide with better information is worth —provided the cost of waiting is bounded. That is why flexibility is worth more, not less, in the turbulent businesses of Part 1.

It changes what failing means. If the pilot was an option premium, the pilot that "fails" and prevents a large mistaken investment is not a failure: it is the option doing its job —McGrath called it "falling forward": the value of an exploration program is not measured by the success rate of its trials but by what it builds and what it avoids. A company that punishes the manager whose honest pilot came back negative is paying insurance premiums and setting fire to the policies.

Jeff Bezos gave the same discipline the names of doors: two-way door decisions (reversible: walk through fast, with a light process) and one-way door decisions (irreversible: there, yes, the full heavy process). The typical organizational failure is the crossover: bureaucratizing the reversible and rushing the irreversible.

Discovery-driven planning: the plan whose output is learning

The tool that turns all of this into an operating method was published by McGrath and MacMillan in 1995, under the exact name of its difference: discovery-driven planning. The premise: on new ground the ratio of assumptions to knowledge is inverted relative to the mature business —so the plan cannot be a forecast to be met; it has to be a device for converting assumptions into knowledge at the lowest possible cost. Its disciplines —four documents in the original article; benchmarking was added in later formulations by the same authors:

1. The reverse income statement. The conventional plan projects revenues and deduces whether the project is worth doing. The reverse one starts from the profit that would make the project worth doing, and works backwards to the revenues, prices, volumes and costs that would produce it. The result is not a forecast: it is the explicit list of what would have to be true —Euro Disney, planned this way, would have had it written down that "this works if the European visitor stays X nights and spends Y"—, with every condition exposed to the question "and who verified this?".

2. Benchmarking against the market. Every assumption in the reverse statement is compared against external references —has anyone, in any market, achieved that spend per customer, that conversion rate, that frequency? It is the outside view of the previous series, institutionalized: assumptions that require breaking world records get flagged in red.

3. The operations specification. Translating the assumptions into concrete operations —how many salespeople, how many visits per sale, what logistics— so that fantasy dies early: many plans collapse not because of the market but because the operation they would require is physically impossible.

4. The assumptions log. The centerpiece: a living, numbered list of every assumption, each one with its owner and its verification date. What the conventional plan hides inside spreadsheet cells, this document puts in plain sight. The cultural rule that goes with it: changing an assumption in the light of new data is not a failure of the plan; it is the plan working.

5. Milestone planning. The budget is not released whole: it is released by milestones, and each milestone is defined by the assumptions it verifies —the market study verifies assumptions 3 and 7; the pilot, 4 and 9; the small plant, 12. Each milestone is a gate where the project can be reinforced, redirected or shut down with a cool head, because the exit conditions were written when nobody's pride was on the line yet (the ex ante exit rules of the previous series, now as a financing system).

In the Hamel and Prahalad volume, McGrath took the idea one step further with an observation large companies tend to miss: an organization's portfolio of experiments is idiosyncratic —it reflects its assets, its customers, its questions— and that is why what it learns from that portfolio is inimitable by construction. Two competitors can copy the same product; they cannot copy the sequence of two hundred trials that taught one of them where not to invest. In turbulence, the defensible advantage moves from the asset to the learning process —which is the fine-grained reconciliation of Ghemawat with Hamel: the irreversible commitment worth making is the commitment to the learning machine.

The dark side of flexibility

This article would be propaganda if it ended here. Flexibility has pathologies of its own, and all three are expensive:

Drift. A company where everything is an option and nothing is a commitment doesn't have a strategy: it has a portfolio of curiosities. Optionality without a thesis to order it —what are we trying to learn, in order to build which position— degenerates into twenty eternal pilots that never scale, each one too small to matter and too beloved to close. A portfolio of options needs what the financial option comes with as standard: an expiration date and a strike price —when the decision gets made, and what signal triggers either the serious investment or the shutdown.

The infinite premium. Keeping doors open costs: management spread thin, blurred brands, sub-scale operations, capital tied up "just in case". Donald Sull called the disciplined version active waiting —keeping reserves and probes while waiting for the big opportunity, and striking with everything when it appears— but he underlined the requirement: waiting is a state of readiness, not a way of life. And Michael Raynor documented the paradox nobody wants to hear: bold-commitment strategies dominate both ends of the distribution —the great successes and the great failures— while flexibility buys survival at the price of giving up the winning extreme. Choosing how much optionality to hold is choosing which part of that distribution to play in; there is no configuration without a cost.

Flexibility in name only. The most common one. The company declares "we work by hypothesis" and goes on rewarding delivery of the number, punishing the negative pilot and automatically renewing last year's projects. Flexibility is neither rhetoric nor a PowerPoint structure: it is a set of decision rights and consequences —who can close what, what gets celebrated, what gets funded by milestone. If the incentive system didn't change, the company didn't become flexible; it became bilingual.

Commit or option: the criterion by type of decision. Ghemawat, 1991; Luehrman, 1998; McGrath & MacMillan, 1995.

Questions for Monday

For your portfolio of projects and investments, exactly as it stands today:

One piece is still missing, the one that holds up the other two. Measuring the regime (Part 1) and managing the portfolio of commitments and options (this part) are decisions taken at the top; but a company's speed of response doesn't live at the top —it lives in its structure: in who can decide what without asking permission. And on that there is a recent empirical result, from several of the authors who measured competent management, that flatly contradicts the instinct of every crisis committee: when the storm arrives, the companies that come through it best are not the ones that concentrate command. Part 3 closes the series there: the organization for turbulence.