The geographic-expansion proposal described 60 venues in the reviewed migration population. That count was accurate. The difficulty was the next assumption: each venue was treated as one planning unit for estimating the work. Discovery found 48 sites with two rooms and 12 with seven. The first group contained 96 rooms and the second 84, for 180 rooms altogether. A correct customer-site count had become an incomplete account of the records and arrangements the company would need to handle.
This is a useful kind of strategy error to examine because it does not begin with an obviously false number. Everyone can agree that there are 60 venues and still disagree about what the number permits the model to assume. The commercial team may think in terms of customer organisations, the implementation team in terms of planning records and the product team in terms of shared capabilities. The problem arises when one unit is quietly asked to represent all of those perspectives.
A legitimate commercial unit can be insufficient for an operating estimate
The expansion team argued that each venue remained one customer and that a common migration approach should absorb the variation. There was truth in that argument. Some work could be shared, and the existence of several rooms did not mean every task had to be performed independently from the beginning. I did not want discovery to become a reason to multiply every cost mechanically. The question was which work varied with the room-level arrangements and which could remain common.
Implementation colleagues could not accept the assumption that all differences would disappear inside a single venue record. Rooms had different availability and production uses, and their planning records needed to retain those distinctions. A model that treated them as interchangeable would make the estimate look simpler without resolving the actual work. We needed a revised account before carrying the same assumption into another geography. That was an operating consequence of the evidence, not a preference for a more complicated spreadsheet.
I supported deferring the ₹3.6 crore geographic-expansion proposal and separately approving ₹1.1 crore for migration redesign. The redesign remains a programme to complete. It should not be described as a successful migration merely because its funding was approved. Product and implementation teams own the detailed work, and customers remain responsible for room availability and their production arrangements. The planning software does not certify a venue's safety or determine whether a production can operate responsibly.
Changing sequence is not the same as realising a saving
The two amounts invited a tempting subtraction. A ₹3.6 crore proposal had been deferred and a ₹1.1 crore programme approved, so the difference could be presented as money saved. That would be misleading. The proposals had different purposes, and the decision changed the sequence of commitments. The expansion might return later with a different case. The redesign still needed to be delivered. Neither fact established a realised financial benefit equal to the arithmetic difference.
I retained the consequence we could actually describe: later market entry and additional work to make the migration commitment more credible. That was less attractive than an account of growth proceeding on schedule. It also avoided asking implementation teams to absorb an understated promise simply because the original model had already made its way into a strategy presentation. A proposal should remain open to correction when discovery changes the work it represents.
The decision also required clarity about what had not changed. Sixty venues remained 60 venues. We had not discovered 180 customers or established that the market was three times larger. The room count was relevant to a particular operating question. It should not be borrowed as a new commercial-growth claim any more than the venue count should be borrowed as a complete estimate of migration work. Different units can coexist when the paper explains the decision each one serves.
I find that distinction valuable when teams disagree. A dispute about which number is right may conceal the fact that both numbers describe legitimate things. The better question is which thing the proposed decision depends on. Once that is clear, the discussion can move from defending a favourite count to understanding the relationship between the counts. That is more constructive than declaring one function commercially naive or another operationally obstructive because they begin with different units.
The model should explain what its unit represents
I now ask for a short explanation of the unit before reviewing the expansion arithmetic. If the model uses a customer, site, room, production or season, what work is assumed to vary with that unit? Which work is common? What evidence supports the relationship? These questions do not replace detailed estimation. They help reveal whether the detail is being applied to a coherent representation of the commitment. Precision in the calculation cannot rescue a unit whose meaning has been left unresolved.
The explanation should also identify where the evidence comes from. Our discovery concerned 60 sites with a particular distribution of rooms. It did not establish the room pattern of every venue in a new geography. A revised model could use the finding to ask better questions, but it should not treat the observed ratio as a universal constant. Expansion into another population needs its own understanding of similarities, differences and the assumptions the company is willing to accept.
That is why I prefer a model that preserves uncertainty over one that offers a single apparently exact answer too early. The strategic decision may still need to be made before every detail is known. The paper can say which assumptions are provisional and which findings would change the recommendation. What it should not do is conceal the uncertainty by giving a convenient commercial label more operating meaning than the evidence supports.
For directors, the case suggests a practical line of enquiry. Ask what the model's unit represents and whether the enterprise has tested that relationship against the work it is promising. Ask how management distinguishes a changed sequence from a realised benefit. Ask which parts of the next programme remain unfinished and who owns them. Those questions help a board examine the basis of an investment without turning it into a migration-design committee.
I also want the decision record to preserve the original assumption. If the final account only describes the improved model, the company loses an opportunity to understand why the shortcut looked reasonable. We had a valid venue count, a plausible belief in shared work and an attractive expansion timetable. The error was allowing those facts to settle the workload question too early. Keeping that sequence visible makes the case useful to the next team facing a similarly persuasive simplification.
Sixty venues and 180 rooms can both be true. Strategy becomes more credible when it explains why the distinction matters to the commitment under consideration. In our case, it changed the timing of expansion and the work funded first. The result was not a neat saving or a completed programme. It was a decision built around a more faithful account of what the company would have to do, and a clearer basis for judging the next proposal when the redesign had progressed.