Why we model three futures
Ask a small business how much runway it has and you get one number. Ask how that number was reached and you usually find one of two mistakes, each the mirror of the other.
The two mistakes
The first is counting the pipeline. Cash on hand, plus everything in conversation, divided by burn. It produces a comfortable figure and it assumes every deal closes, which no pipeline has ever done. The number is not a forecast; it is a hope with arithmetic applied.
The second is refusing to count it. Cash on hand divided by burn, full stop. This is defensible and it is what a cautious person does, but it produces a figure so far below what anyone expects that nobody uses it to decide anything. A plan people ignore is not a conservative plan. It is no plan.
Both produce one number. Both are wrong in a direction the number itself does not disclose.
What we do instead
Every income entry that is not contracted carries a percentage. The model then reports three figures from the same inputs:
- Pessimistic, counting only what is certain. This is the floor and it is
the number that answers "what if nothing new closes".
- Expected, with each uncertain entry weighted by its probability. This is
the planning number.
- Optimistic, counting everything at face value. This is the ceiling, and
it is shown mainly so that nobody mistakes the expected figure for it.
Three numbers take longer to read than one. They are also the only honest summary of what is actually known, because the spread between them is the uncertainty, and a single figure deletes it.
The part that changed how we use it
Three fixed scenarios turned out not to be enough, and the reason is worth recording because it was not obvious at the start.
Nobody models one future. They model the one they fear and the one they are hoping for, and those two are rarely the presets. The useful question is not "what does expected say", it is "what does expected say without the largest retainer", and that question needs a scenario that can be forked, renamed and kept alongside the original.
So the three ship as starting points rather than as the ceiling. A scenario can be duplicated, edited and compared against any other on the same chart. The what-if controls apply to whichever scenario is selected and never quietly overwrite a saved one, because a planning tool that loses the version you were comparing against is worse than no tool.
Why probability beats padding
The alternative to a percentage is padding: add a margin to every line and call it prudence. It is easier, and it is worse for two reasons.
Padding is invisible. Once it is inside the numbers, nobody can tell how much of the plan is real and how much is cushion, and the cushion compounds every time someone adds their own.
Padding is also uniform, and risk is not. A deal with a 20 percent chance and a deal with an 80 percent chance are different facts, and a flat margin over both treats them as the same one.
A probability stays visible, stays per line, and can be argued with. Somebody can look at 60 percent and say "that is a 30", which is a conversation worth having. Nobody has ever usefully argued with a cushion.
What it still does not tell you
The probabilities are guesses. Weighting a guess does not make it a measurement, and the expected figure carries every optimism in the inputs straight through to the output.
What the three futures buy is not accuracy. It is that the optimism is in a column somebody can point at, instead of dissolved into a single number that looks like a fact.