In a due diligence, consultants check hundreds of assumptions. In the end, three to five of them matter for the purchase decision. The rest mostly costs one thing: expensive hours.
The effort is understandable, admittedly. If you buy a company, you want the data rooms analyzed and the contracts reviewed, and one wrong assumption can cost millions. The problem is that nobody knows beforehand which assumptions actually move the result. So everything gets checked with the same thoroughness.
Don't check every number, check the few that the decision really depends on.
A number says nothing about how certain it is
An example. A mid-sized machine builder wants to acquire a software company for 20 million euros. The software helps industrial customers maintain their machines, and after the acquisition it is supposed to be sold to the machine builder's own customers as well.
Anna, the CFO, builds the business case with her team: 10,000 of their own customers are potential buyers, 4 percent of them buy, and each brings in 2,000 euros of revenue per year. Together with existing customers, purchase price, costs and churn, this adds up to a net present value (NPV) of five million euros, which, simply put, is how much additional value the acquisition is expected to create.
But the assumptions are not equally reliable. The company knows the number of its own customers fairly precisely. The 4 percent purchase rate, on the other hand, is just an estimate from sales; the software has never been offered to these customers, and the actual rate could just as well be 1 or 8 percent. In Excel, both assumptions still look the same: a neatly entered number.
Step 1: Ranges instead of point values
Anna replaces the estimates with realistic ranges. Instead of exactly 10,000 customers: 9,500 to 10,500. For the purchase rate, where the uncertainty is much larger: 1 to 8 percent. The same for price, costs and the remaining assumptions.
This shows exactly what the company knows and where the gaps are. A single number hides that information.
Step 2: Calculating the possible outcomes
Instead of calculating the case once with fixed numbers, it is now calculated many thousands of times, each time with a different combination drawn from the ranges. In one run, 2 percent of customers buy, in another 6, while price and costs change at the same time. This method is called Monte Carlo simulation.
The result: the net present value is not reliably five million euros, it lies somewhere between minus one and plus twelve million. At minus one million the company would destroy value, at twelve the acquisition would be very attractive. The five million from the first calculation are just one of many possible outcomes.
This doesn't tell Anna yet what will happen. But for the first time, she sees how large the uncertainty really is.
Step 3: Finding the critical assumptions
This range is too wide to approve the acquisition with a clear conscience. Where does it come from? A sensitivity analysis shows how strongly the uncertainty of each individual assumption moves the overall result.
The result is clear. Whether the machine builder has 9,500 or 10,500 suitable customers changes little about the range of the net present value, and the same goes for deviations in costs. The purchase rate, on the other hand, decides it: at the lower end the acquisition becomes a loss-making deal, at the upper end it creates considerable value.
The purchase rate is part of the “critical few”, the few assumptions the decision actually depends on. The other numbers simply don't deserve the same checking effort at this point.
Step 4: Checking in a targeted way
Instead of spending more weeks checking every Excel cell, Anna and the consultants focus on the purchase rate. They talk to customers, test the offer with a small group, and look at benchmarks from similar products. After that, the rate can be narrowed down: 3 to 5 percent instead of 1 to 8.
Recalculated, the net present value now lies between three and seven million euros. Even in the cautious scenario it stays positive. Management can approve without having to rely on a seemingly exact number.
If the range were still too wide, the second most important assumption would be next, and so on until the remaining uncertainty is acceptable for the decision. And a checked assumption is allowed to remain a range. Nobody can guarantee that exactly 4 percent of customers will buy.
Not every number deserves the same attention
A due diligence doesn't get better because every assumption was checked with the same effort. What counts is whether the remaining uncertainty still affects the purchase decision. If you measure that, you can see where additional information really helps and where further analysis only costs time and consulting fees.
The actual takeaway is simple: don't check every number, check the few that the decision really depends on. That is exactly what Bayescase supports out of the box: from honest ranges through the simulation to identifying the critical assumption.
