Stress-Testing a Plan With Three Scenarios

Why this matters

Most expansion plans in a small shop are built on one forecast: the one where things go about as hoped. That forecast is usually reasonable, and it is still the wrong thing to decide on, because the plan does not kill a shop in the average year. It kills it in the year the new crew never fills up, the weather turns, and the equipment payments keep arriving. Three scenarios, each carried to its lowest cash point, take an afternoon. They tell you whether the bad year is survivable, and, more usefully, what you will do in it, decided while you can still think straight.

The plan we will test

A lawn and landscape shop runs two mowing crews and wants a third for the coming season. The commitment:

• A truck, trailer and commercial mower package, financed, with a down payment at signing in late February.
• A crew lead and a crew member hired for March, kept on through the winter to hold onto them.
• Target: a full route of residential and light commercial mowing from May.

Every money figure below is in one unit: crew-months (CM), meaning one month of the new crew's fully loaded cost (two people's wages and payroll costs, the equipment payment, fuel and insurance). The shop has set aside a cash reserve of 4.0 CM for the expansion; the existing two crews are assumed to run at break-even over the year, so the question is purely what the third crew does to that reserve. Runway at any point means how many months the remaining reserve would cover the crew's net monthly loss at that time.

Step 1: Write the commitment with its fixed costs separated

List what you owe whether or not the plan works, and when. Here: a down payment of 1.0 CM at signing, then the equipment payment and insurance every month regardless of how many lawns get cut, and two people's wages every month you keep them.

Why first: fixed costs are what turn a slow month into a cash hole. Revenue moves with the scenario; these do not. What breaks if you skip it: the bad case gets modeled as "less revenue" while costs quietly shrink along with it in your head, and the scenario looks gentler than the contract you signed.

After the down payment, the reserve stands at 4.0 minus 1.0, 3.0 CM.

Step 2: Build the base case from your own history, not your hopes

The base case is what happened when your first two crews launched, adjusted for anything genuinely different. It is not the target. The owner's records from the second crew's first year give the shape:

• March to April (2 months): half-booked while the route fills. Net -0.5 CM a month.
• May to September (5 months): full route. Net +0.4 CM a month.
• October to November (2 months): leaf cleanups and aeration. Net +0.1 CM a month.
• December to February (3 months): almost no work, crew kept on. Net -0.8 CM a month.

Walking the reserve through the year: 3.0, minus 1.0 for spring, is 2.0. Plus 2.0 for summer is 4.0. Plus 0.2 for fall is 4.2. Minus 2.4 for winter is 1.8 CM at the end of February.

Why your own history: an expansion forecast built on hope runs optimistic in a predictable direction. Using the record of something you have already done once is the cheapest outside view available; see related: Forecasting a New Service Line From a Reference Class. What breaks if you skip it: the base case becomes the plan's sales pitch, and the "bad" case is just the base case with a haircut.

Note what the base case already shows: the reserve peaks at 4.2 in November and falls 2.4 over the winter. Anyone who looked at the shop's bank account in October would think the expansion was a clear success, and the lowest point of the year is still ahead.

Step 3: Build the bad case from something that has already happened

The bad case is not a disaster. It is one thing going wrong that has gone wrong before, in this shop or a neighbor's. The owner picks the drought summer of three years ago, when watering restrictions cut mowing frequency and customers skipped visits. In that year, summer contribution per crew fell to about a quarter of normal.

So the bad case changes one line: May to September at +0.1 CM a month instead of +0.4. Everything else stays as the base case.

Walking it: 3.0, minus 1.0, is 2.0. Plus 0.5 for summer is 2.5. Plus 0.2 is 2.7. Minus 2.4 is 0.3 CM at the end of February. At the winter loss of 0.8 CM a month, that is about 0.4 months of runway (0.3 divided by 0.8) at the low point.

Why "has happened before": it stops the argument about whether the bad case is realistic. Nobody can say a drought is far-fetched in a shop that lived through one. What breaks if you skip it: the bad case gets chosen to be survivable, which defeats the purpose.

Step 4: Build the worse case from two things that tend to arrive together

The worse case combines two problems, and the useful ones are correlated: problems that make each other more likely, or share a cause. A drought is bad. A drought in a year when you also cannot hire, because every landscaper in town is short-staffed, is a realistic pairing, not bad luck stacked on bad luck.

Here: drought summer, plus no crew lead found until May. The equipment payments and insurance start in March anyway, and the crew member hired early does whatever work the other crews can spare. Spring becomes -0.8 CM a month instead of -0.5.

Walking it: 3.0, minus 1.6 for spring, is 1.4. Plus 0.5 for the drought summer is 1.9. Plus 0.2 is 2.1. Then winter, month by month: December takes it to 1.3, January to 0.5, and February to -0.3 CM. The reserve runs out a little past the middle of February (0.5 divided by 0.8 is about 0.6 of a month).

All figures in CM, end of each phase.

What breaks if you skip it: you test each risk separately, each one looks survivable, and the year they arrive together is the year you did not model.

Step 5: Decide now what you would do in each case

This is the step that makes the exercise worth an afternoon. For each scenario, write the signal that would tell you you are in it, the date you will check for that signal, and the action you will take. Decided in February, these are commitments. Decided in December with 0.5 CM left, they are panic.

The owner's three:

• Signal of the bad case: by June 15, the third crew's route is booked under 70% of capacity. Action: the third crew is laid off seasonally from December 1 (with an offer to rehire in March), cutting the winter loss from 0.8 to 0.3 CM a month, which leaves only the equipment payment and insurance.
• Signal of the worse case: no crew lead hired by March 15. Action: the crew member is folded into the existing crews, where he is billable, and the third truck goes on light duty until a lead is found.
• Signal of the base case: neither of the above. Action: keep the crew on all winter as planned.

Re-run the bad case with its action: winter becomes three months at -0.3, or -0.9, so the end of February is 2.7 minus 0.9, 1.8 CM. Re-run the worse case with both actions: if folding the crew member into the other crews cuts the spring loss from -0.8 back to about -0.6 a month, spring costs 1.2 instead of 1.6, and the reserve goes 3.0, 1.8, 2.3, 2.5, then minus 0.9 for winter, 1.6 CM at the end of February. A shop that was going to run out of money in February now ends the year with more than a third of its reserve, using decisions it could have made in a single meeting.

What breaks if you skip it: the scenarios become a report rather than a plan. Owners who model the bad case and do not decide the response usually recognize the bad case only after the action that would have helped is no longer available. Setting the signal and date in advance is the same discipline as kill criteria; see related: Setting Kill Criteria Before You Start Something New.

Step 6: Check the worse case against a hard floor, and change the plan if it fails

Pick one number the worse case must not cross, with the actions included. For most shops it is a minimum reserve you never plan to dip below, because something unmodeled always happens too. If the worse case with actions still crosses it, the answer is not to stop worrying. It is to change the commitment.

The owner set the floor at 1.0 CM, one full month of the new crew's cost still in the bank. The worse case with actions ended at 1.6, so the plan passed. But the exercise had already changed it: since the worse case was driven by hiring, the owner moved the sequence, hiring the lead first and signing for the equipment only once the lead had accepted. That one change makes the worse case's March and April payments disappear, which no amount of trying harder in June could have done.

What changes the answer

A commitment you can exit. If the equipment were leased month to month, or rented for the first season, the worse case loses most of its fixed cost and the floor check matters much less. The more reversible the commitment, the less an afternoon of scenarios is worth; see related: Reversible and Irreversible Decisions in a Small Shop.

A shop already near its floor. If the reserve before the expansion were 2.0 CM rather than 4.0, the same moves shift every line down by 2.0: the bad case without action ends at -1.7, and even with its action at -0.2, under the 1.0 floor. Then the plan waits a year, or shrinks.

Scenarios mistaken for forecasts. None of the three is a prediction. You do not need to know how likely the worse case is. You need to know you would survive it and what you would do.

References

• See related: Running a Premortem Before a Big Commitment, which is a good way to generate the bad and worse cases
• See related: Forecasting a New Service Line From a Reference Class
• See related: Setting Kill Criteria Before You Start Something New
• See related: Scarcity and the Decisions a Cash Crunch Makes
• See related: Cash vs Profit: Why They're Different