Most FP&A teams are experts at reporting what happened, but rely heavily on static spreadsheets and negotiation to guess what’s next. Joshua Zable, President & CFO of Minitab, explores why overcoming 'statistics anxiety' and using basic quantitative baselines is the ultimate sanity check for enterprise decision-making.
Walk into almost any finance department and ask how the annual budget or quarterly forecast is produced, and you’ll likely hear the same answer: spreadsheets, management judgment, and a series of meetings where business leaders negotiate revenue and expense assumptions.
Despite the extraordinary advances in analytics over the past several decades, many organizations still forecast the future using methods that have changed surprisingly little. Ironically, while finance organizations have become experts at reporting what happened, many have invested far less in scientifically predicting what will happen next.
As CFOs, we pride ourselves on being data-driven. Yet when it comes to forecasting, many organizations rely more on intuition than statistical evidence.
That needs to change.
It’s not surprising though. The typical path to becoming CFO runs through accounting, auditing, FP&A, corporate finance, treasury, or investment banking. Surveys of finance leaders consistently show that many CFOs began their careers as CPAs, controllers, auditors, or financial analysts before advancing into executive leadership. Business school curricula often require economics and finance, but only limited coursework in statistics beyond introductory quantitative methods.
Every capital investment, hiring decision, acquisition, pricing strategy, and cash flow plan depends on an estimate of the future.
If forecasting is one of our most important responsibilities, shouldn’t we be using the most scientifically sound methods available?
Despite being the self-proclaimed Statistical CFO, I’ll be the first one to tell you that statistics alone don’t make you an excellent forecaster. I’m not suggesting you replace all the things you’re doing today. Business understanding and experience are critical to making good predictions. But trend analysis? Do you really believe eyeballing and assigning growth rates are more scientifically sound than the statistical methods invented to solve this very problem?
I’ll tell you why. First of all, statistics anxiety is a real thing. Studies have proven that students have real anxiety about learning statistics, with the most referenced study saying that it impacts 80% of them. While many of us outgrow things that gave us anxiety in college – most of us don’t have to confront our statistical fears. For nearly 20 years I worked analytical jobs on Wall Street and Main Street and never had to dust off the old statistics textbook. It was only when I met the brilliant (and patient!) statisticians working at my current company, that I felt comfortable enough to dive back into statistics.
Second, there isn’t one statistical method for forecasting. And while that can seem overwhelming, the different statistical methods can help financial leaders account for the various things in your head. Picking the proper method can crystallize how you’re thinking of forecasting. Worried about seasonality? No problem. Want to put more weight on recent results because the business has changed? There’s an algorithm for that. Are you a start up experiencing a massive sales ramp? There’s a model for that.
Naturally, an algorithm can’t account for everything. You may be ramping marketing and expect sales to accelerate. Something may have shifted in your industry – for better or worse. However, an algorithm is the closest thing you have to a sanity check (don’t tell your Board I said that). Put simply, if a statistical algorithm is providing a response significantly different than your forecast, it challenges you to answer the why. Or else – assuming you’ve selected the right method – you’re more than likely going to get a more accurate prediction than the one you have.
Begin by clearly defining what you want to forecast and the decision this forecast will inform. A well-defined business question makes it easier to identify the necessary data and analytical approach.
Use a straightforward statistical or forecasting tool to start, such as forecasting an existing trend. Most analytics platforms can help you select an appropriate method, and AI tools can clarify available options and their assumptions.
If your financial system includes forecasting features, explore them as well. However, do not treat the output as a black box. Understand the algorithm or methodology used and assess its suitability for your business.
Compare the statistical forecast with your traditional forecast. You may choose to use the statistical forecast directly, include it in a range of outcomes, or recognize that significant business changes may limit the usefulness of historical patterns.
The goal is not to replace financial judgment with statistics, but to make that judgment better informed.
As you incorporate statistical thinking into your forecasting, you will gain confidence in recognizing when data provides new insights and when experience and industry context should guide decisions. You may even impress a Board member.
More importantly, you will strengthen your team’s ability to combine financial expertise, business judgment, and mathematical rigor to produce better forecasts and make more informed decisions.
As you add statistics to your forecasting repertoire, you’ll gain confidence and differentiate yourself. Heck, you may even impress a Board member or two. But most importantly, you’ll be adding critical skills to make your forecasting better and therefore more likely to be accurate.
