With an uncertain business case, a CFO wants to know two things: why it makes sense to spend money now, and how you will notice in time that you're wrong. For that you don't need a certain final number. You need the one assumption that carries the case, an amount to test it and a clear recommendation for what happens next.
What does the CFO ask when the numbers are uncertain?
Take as an example an automation pilot that an innovation manager brings to the steering committee. Technically the whole thing works, employees save time, and the business case looks good. Then the CFO asks: “Okay, people save time. But what happens with that time? Can we handle more orders because of it? Do we avoid new hires? Where does the economic effect come from?”
That is exactly what hasn't been checked yet. But the savings are already in the business case, as if they were an economic effect already. That's where it gets difficult.
Does it help to calculate conservatively when the numbers are uncertain?
The usual advice for this situation is: calculate conservatively, show scenarios, speak the CFO's language. ChatGPT and Claude basically answer the same way when you ask them. “Calculate conservatively” sounds reasonable at first. But an uncertain assumption doesn't become more reliable just because I make the number smaller. If in doubt, we make a good idea look bad without knowing any more afterwards.
An uncertain assumption doesn't become more reliable just because I make the number smaller.
Three scenarios don't help much either if nobody can explain why one of them should be more likely or what we conclude from it. In the example, the worst case then simply assumes fewer hours saved, and the CFO's question about what happens with that time is just as open as before.
What do you already know, and what are you only assuming?
For me, this advice misses the actual question: What do we already know, what are we only assuming, and is that enough to fund the next step? In the automation pilot, the team knows that the technology works and that time is saved. That this time turns into money, through more orders or avoided hires, is so far only an assumption. And exactly this assumption carries the case.
At Bayescase, innovation management software, we make these load-bearing assumptions visible and put a number on what they mean for a project's impact and risk. Which assumptions deserve checking in the first place, we worked through with an example.
Once it's clear which assumption carries the case, you have the answer to what the CFO is really checking. You need to be able to explain how you will test this assumption and what happens if it doesn't hold. In the example, that means: How do you show whether the time saved really leads to more orders or fewer new hires? What does this test cost? And from when on will you know?
Don't committees want a number after all?
An objection you often hear at this point: committees want a number, not a range. There is something to that. A committee needs a concrete recommendation, and just putting down a range and saying “take your pick” isn't enough. The range is more honest than the single number, but it doesn't release me from recommending a decision.
As long as the load-bearing assumption is still being tested, I don't know exactly what will come out at the end. I can, however, say very concretely how much money I need for the next step. Those are two different numbers that tend to get mixed up in these discussions: the amount you want approved now, and the result the project is supposed to deliver at some point. You can state the first one precisely; the second stays a range for now.
What you should write down before the next budget round
Because the committee still needs a recommendation in the end, write this sentence down once before the next budget round and fill it in for your project: “I recommend approving this amount now, because we already know the following. With it, we test this open assumption. Depending on the result, we continue, adjust the project or stop.”
If you can't fill it in concretely yet, you know what to work on before the meeting.
Frequently asked questions
- How do I convince the CFO of a business case when the numbers are uncertain?
- Show them why it makes sense to spend money now, and how you will notice in time that you're wrong. To do that, separate what you already know from what you are only assuming, name the assumption that carries the case, and recommend a concrete amount to test it.
- Should I calculate an uncertain business case conservatively?
- Calculating conservatively sounds reasonable, but it doesn't make an uncertain assumption more reliable. If in doubt, you make a good idea look bad without knowing any more afterwards. It helps more to name the load-bearing assumption and say how you will test it.
- Do committees want to see a number or a range?
- A committee needs a concrete recommendation; a range alone isn't enough. But you can state the amount for the next step precisely, even if the final result is still uncertain. Those are two different numbers that you should keep apart in the discussion.
- How do I find the assumption that carries my business case?
- For each assumption, ask whether you already know it or are only assuming it, and how strongly it moves the result. Usually the decision depends on three to five assumptions. Bayescase, innovation management software, calculates impact and risk with ranges and uses a sensitivity analysis to show which assumptions move the result the most.
