Forecasting the Future: An Imperfect Science

Today we live in a world where managers and shareholders expect you to meet the forecast within 5% or you may suffer the consequences. However, a forecast is just that: a forecast, not a perfect prediction or even close to perfect prediction. If somebody consistently meets their forecast quarter after quarter, year after year, then you must seriously consider that the earnings are being managed or worse. You might be okay with managed earnings but they often skirt the edge of the unethical and can quickly slide into the illegal.

And dipping into the illegal never ends well.

But there are ways of improving your forecasting. I’m going to list some of the things that can help improve forecasts but won’t ensure perfect predictions. I am not going to address the unrealistic expectations as I do not have the answer for that other than the CEOs will have to work on the shareholders’ expectations.

It helps to have great managers who can intuitively tell where the numbers will fall. I’ve been fortunate enough to have such managers with an incredible feel for the direction of the business that, at least for a year, my forecasts, especially revenue, would hit within $1000 variance each month (in a large program with at least $1 million in revenues per month). I didn’t do the revenue booking and the accountant did not receive my forecast so there was no managed accounting.

Aside from having a great manager, here are a few ideas on improving your forecasts:

1. Prepare by looking at the history first but keep in mind that history is often not predictive of the future. A look at the history may clue you in to “seasonal” trends that you have to take into consideration when doing the forecast.

2. If your operations team have meetings or some kind of lunch and learn, attend them. You can learn about what they are thinking and possibly what the plans are for the future. The more you understand their approach, the better you will be able to think through your assumptions when developing the forecast. Please keep a log of your assumptions.

3. Develop a “starter” forecast instead of dumping the work on the operation folks. This tactic may make you look like a star with operations management. If you develop a good chunk of the forecast before going to the operations folks, you will have taken most of the work off their shoulders. They in turn may regard you in a different and more positive light. I once worked on a team where I developed most of the forecast so all that was left was to discuss it and see what needed tweaking. While I regarded that as a normal thing to do, the operations folks regarded it as proactive. Apparently the rest of my peers were not so proactive and therefore, I differentiated myself.

4. Don’t forecast at very granular levels. You want to break out the broad categories of expenses or revenues because you will never be able to forecast each “general ledger account” that accurately. Early in my career, I once did an experiment where I forecasted for each GL account and then rolled up the numbers and found that the top level number did not make sense. So make a judgment call and figure out the level of granularity that will aid you in developing the forecast.

5. For programs or business units where the labor headcount does not fluctuate from month to month, don’t straight line the labor cost but instead take into account holidays and, if possible, vacation time. These time offs can greatly swing your forecast so by making an effort to eliminate labor as one of the items causing your variance, you get can closer to the 5% variance threshold (5% being the typical variance threshold that I have encountered). There are a lot of other things that can mess you up so try to eliminate labor as one of your variances (again, you won’t be perfect, but at least you are trying to narrow the swings).

All of these suggestions, except maybe item 4, seem like common sense but I have learned that most people don’t apply common sense. Instead they take short cuts to do the work. If you do at least the five aforementioned tactics, your forecasts should improve.