पाठशाला Pathshala · हिसाब Hisāb, Unit economics · Lesson 30 · Scale
Scenario planning: the three cases and what breaks each
Base, bear and bull cases earn their place only when built from the same few drivers and tested one at a time. How to find the assumption that moves the outcome most and the combination that breaks the company.
Pathshala, The Founder Library · 11 October 2026 · 7 min read
Most startup scenarios are three copies of the same spreadsheet with the growth rate changed: 10 per cent, 20 per cent, 30 per cent. They look rigorous and teach nothing, because they move only the number the founder already hopes for and leave alone the ones that actually decide whether the company survives. Three cases earn their place when they answer two questions: which single assumption moves the outcome most, and which combination of bad news ends the company.
This lesson sets out why three cases, how to choose the drivers, how to set bear and bull values honestly, works one company through, shows how to rank the drivers, and ends with the quarterly review that keeps the cases alive.
Why three cases and not one
A single forecast is a statement of hope with a date on it. Pierre Wack, who built scenario planning at Royal Dutch Shell, opened his 1985 Harvard Business Review article Scenarios: Uncharted Waters Ahead by noting that forecasting errors had become more frequent and sometimes dramatic since the early 1970s, yet companies kept forecasting because nobody had found a better way to handle uncertainty. Scenarios were his answer: not a better prediction, but a set of plausible futures a management team has thought through in advance.
A startup needs the small version. The base case is what happens if the company keeps doing what it has been doing. The bear case is what happens if the things it worries about come true together. The bull case is what happens if the bets pay off. The point is not to choose one. It is to know, before the quarter begins, what the company would do in each, and what signal would tell it which one it is in. Sequoia ran a forecasting and scenario planning session for its founders in May 2022 for exactly this reason: to help them adjust plans to a market that had changed faster than their forecasts.
Choosing the drivers
Use the same model for all three cases, built on drivers rather than on revenue, as the [one-spreadsheet model](/library/one-spreadsheet-model-every-founder-should-build) is. Four or five drivers are enough for most companies: monthly revenue growth, gross margin, the growth rate of operating cost, days to collect, and for subscription businesses churn or for paid-acquisition businesses CAC. Each should be a number the company measures every month, so that the cases can be checked against reality.
Resist adding drivers. Each additional one makes the cases harder to read and the tornado harder to trust. If a sixth driver matters, it will show up as a large gap between forecast and actual in one of the five, and can be split out then.
Indian companies should look hard at two drivers that global templates treat lightly. Days to collect move sharply with customer mix: a company adding large enterprise or government customers can see them double within a year. And costs billed in dollars, cloud and software subscriptions above all, move with the exchange rate rather than with the company’s decisions. Where either is a large share of the business, give it its own driver rather than burying it in margin or cost growth.
Setting the bear and the bull honestly
The base case uses what the company has actually achieved over the last three to six months, not the plan. The bear and bull values come from the company’s own history first: its worst and best quarters for each driver in the last two years. Where there is no history, use a fixed distance from base and say so: two points of monthly growth either way, seven points of gross margin, one point of monthly cost growth, thirty days of collection.
Two errors are common. The first is a bear case that is merely a slower base, with growth cut and everything else unchanged; real bad quarters arrive in clusters, slow sales with longer collections and a margin squeezed by discounting. The second is a bull case with no cost: faster growth usually needs more people, more inventory or more credit to customers, and the model should carry it. A useful check is Gary Klein’s premortem: assume the plan has failed and ask the team to write down why. The reasons that come up most often belong in the bear case.
One company, three cases
A Hyderabad company sells compliance software to mid-sized manufacturers. It makes ₹40 lakh of revenue a month at a 60 per cent gross margin, spends ₹40 lakh a month on operating cost, collects in 45 days and has ₹5 crore in the bank. Its base case is the last six months repeated: revenue growing 4 per cent a month, cost growing 1.5 per cent.
In the base case it reaches operating breakeven in month 21 and holds ₹2.2 crore at month 24. In the bear case, growth of 2 per cent, a 53 per cent margin, cost growing 2.5 per cent and 75 days to collect, the cash runs out in month 19. In the bull case it breaks even in month 8 and holds ₹8.6 crore. The range is wide, as it should be. The useful work starts now.
Finding the assumption that moves the outcome most
Swing each driver alone from its bear value to its bull value, with the others at base, and record month-24 cash. Sort the drivers by the width of the swing. That chart, a tornado for its shape, is the most useful single picture in a plan. Here growth sits on top, moving month-24 cash from ₹46 lakh to ₹4.5 crore; cost growth comes next at ₹57 lakh to ₹3.6 crore; then gross margin at ₹1.1 crore to ₹3.3 crore; and days to collect last at ₹1.6 crore to ₹2.8 crore.
The ranking says where management attention belongs. Growth is the assumption to watch weekly and the one the next hire or experiment should serve. Cost growth, the second, is almost entirely a hiring decision and so within the founders’ control. Collections, which consume much of the finance team’s worry, matter least here. A different company will rank them differently, which is why the chart has to be drawn rather than assumed.
The bear case does not kill a company. A particular combination of bear values does, and it can be named in advance.
What breaks each case
No single bear value runs this company out of cash in 24 months. Pairs do. Bear growth with bear cost growth ends the cash in month 21. Bear margin with bear cost growth ends it in month 22. Bear growth with bear margin ends it in month 23. Bear growth with slow collections survives, with ₹22 lakh left. That is the real output of the exercise: the company can absorb any one bad assumption and any two of growth, margin and cost will break it.
Each break gets a trigger and an action, in the form the [path-to-profitability memo](/library/path-to-profitability-memo) uses. If growth runs below 3 per cent for two consecutive months, cost growth goes to zero: no new hires until it recovers. If gross margin falls below 56 per cent, the infrastructure and discounting levers are pulled that month. The bull case gets triggers too: if growth runs above 5 per cent for a quarter, the company should know in advance which hires it will bring forward. Paul Graham’s Default Alive or Default Dead? argues that the danger lies in starting to worry too late; a trigger set in advance moves the worry to the right month.
The breaking combinations also size the cushion a company should hold. If two bear values together exhaust the cash in month 21, the company needs either enough extra cash to survive that combination for another two quarters, or triggers that act early enough to stop it reaching month 21. Raising money is the slower of the two and should start at least six months before the month the pair would break the company.
The quarterly scenario review
Once a quarter, after the books close, reset the base case to the last six months of actuals and re-run all three cases. Redraw the tornado and note whether the top driver changed. Re-test the pairs and note which combinations now break the company and in which month. Check the triggers against the quarter that has just ended and record which, if any, fired.
Put one page in the board pack: the three cases with month-24 cash and breakeven month, the tornado, the breaking combinations and the triggers. Each month in between, mark the actuals for each driver against its bear, base and bull values; a driver tracking its bear value for two months is a signal to act before the quarter ends. A company that does this for a year stops arguing about whether the forecast is right and starts arguing about which driver to move, which is the argument worth having.
Nothing here is legal, tax or investment advice. The Hyderabad company is illustrative; its figures are the figure’s defaults so that every number in the text can be reproduced.
Sources
- Pierre Wack, Scenarios: Uncharted Waters Ahead, Harvard Business Review, September 1985 — Forecasting errors more frequent since the early 1970s; the Shell approach to scenarios.
- Gary Klein, Performing a Project Premortem, Harvard Business Review, September 2007 — Assume the project has failed and generate plausible reasons for its demise.
- Ravi Gupta and Pat Grady, Forecasting & Scenario Planning, Sequoia Capital, June 2022 — Session for Sequoia founders in May 2022 on adjusting forecasts to a changed environment.
- Paul Graham, Default Alive or Default Dead?, October 2015 — The risk of starting to worry too late; keep a written plan B.