There is an old saying that “Numbers don’t lie”. Unfortunately, the assumptions used to generate them do! In todays “data analytics” driven world, this is an important issue to keep in mind. One of the most important things I had to learn as an early career manager was the lack of value that a financial model provides when decoupled from the assumptions it is based on. Really this goes for any model (at least any model of a real world issue). The real value of most predictive models is actually in identifying the assumptions and clarifying their effects rather than providing a specific quantitative result. Much too often I have seen business plans and pro forma presented with vague assumptions that are discussed only in the small print or not at all. The primary discussion should actually center on the assumptions used to generate the financial model and how those assumptions will be tested and clarified moving forward. Without a very clear understanding of the underlying assumptions, a financial model just gives a false sense of security. In most situations, subtle changes to some of the assumptions can provide results that change a decision from a go to a no-go. For Master of Engineering Management (MEM) graduates with strong quantitative backgrounds, it is especially important not to rely too heavily on the numbers generated by an impressive, complex model you have developed. Rather, as an early career employee, one of your jobs should be highlighting the assumptions for decision-makers and clarifying where the assumptions are just educated guesses and where they have a high probability of being accurate. Be sure not to hide behind the numbers generated by an impressive model. And follow-up with a plan for how the risk of a bad assumption will be managed and minimized over time. As a manager and decision maker, a significant part of the job is how to ask the right questions to ferret out bad assumptions or areas where the assumptions were implicit but not even identified. After all, if most situations were deterministic rather than probabilistic, gathering more information would make decisions easy and judgement would not be so important, just run a model for the answer!
A Blog by Greg Satell, Digital Toronto, has an excellent and more thorough discussion of “How Numbers Lie”. I highly recommend it (along with many posts that Greg has written):