Your Forecast Will Always Be Wrong. And That's Okay.
- ryanbiesecker
- Mar 23, 2023
- 7 min read
Updated: Apr 17, 2023
Setting realistic expectations when it comes to predicting key performance indicators at your place of work.
By Ryan Biesecker
Forecasting and planning are two essential and incredibly challenging functions at nearly any business. Without a forecast to compare against, how does management know if last week's sales were successful or not? I have spent years helping companies big and small run forecasts to support revenue and sales functions and can confidently tell you I've never been 100% accurate, nor have I ever expected to.
Here are a few tricks that don't have to do with the nuts and bolts of the actual predictive math, but instead about managing vertical and lateral forecast reports and communications.
Setting More Realistic Expectations
In this post, I'm going to cover some of the basics of forecasting and planning by answering several key questions below. Take these, along with some spreadsheets and mathematical know-how, to become a bedrock of reliability at your company.
Why Are Forecasting And Planning So Important?
Forecasting sounds like something that should be left to the weatherman, right? I would instead argue that every savvy executive likely reviews forecasts on at least a weekly basis. Those executives' forecasting and planning teams develop tools and reports that predict the future of the business. The basic role of any prediction is to set one or more expectations.
As we know from decades of psychological studies, expectations impact our behavior, or in other words, how we respond to business performance. I guarantee that your boss will look at a final month's sales of double-digit losses very differently if he knew it was coming a few weeks ahead of time.
The further in advance leaders and stakeholders know about changes in key performance indicators (KPIs), the sooner they can start acting on them. If they know far enough ahead of time, effective leaders can execute responses that not only stave off downturns, but turn them into wins.
What's The Difference Between A Forecast And A Plan?
Anyone who tells you forecasting is pure science is probably selling you something that won't paint a complete picture. I'm here to tell you that like most applications of data analysis, running predictions for a business is a delicate balance of art and science.
Let's set some definitions so that everyone is on the same page. A forecast is the explicitly prediction one or more KPIs over a set period of time using historical data, math, and/or assumptions. This is the science side of the equation, built from the bottom up with simple modeling tools or complex statistical simulators to drive numeric outcomes with the goal of trying to be as accurate as possible. Typically, forecasts are for shorter periods of time in the near future. For example "Next week's sales will be $12,500", or "February will have an average spend of $10.70 per customer."
Robust forecasts tend to predict many interconnected, specific KPIs that impact various stakeholder teams, but not necessarily every member of the C-suite. A Chief Marketing Officer at a theme park group might not need to care about the % of warehouse inventory taken up by refrigerated goods each month, but sure cares about weekly gross sales volume and revenue. Conversely, a Chief Operations Officer might want to know about those refrigerated product figures, while they're maybe not as interested in the differences in center sales between the east and west zones of the country. All the metrics matter, but knowing your audience is equally important.
A plan, sometimes also called a "goal" or "budget", is exactly what it sounds like: a high-level answer to the question "how do we want business to do" for a longer time period. Business plans, like personal goals, should be high but achievable. This is the art side of the equation. Plans are typically set by senior leadership to push their teams to the max, and often involve a lot of negotiation and risk assessment. Planning does not usually cover every KPI until the final answer is set, for example "$10M in sales in 2023" or "EBITA of 12% by 2028". Then a planning team will work backwards from the answer to prove out "how to get there".
In the lifecycle of a healthy business year, a plan is set for a handful of KPIs well before the year starts. Then, about one quarter before the year starts, the forecasting and planning team should build out a detailed breakdown of all of the KPIs that eventually lead to the planned final answer. Once set before the start of the year, a plan should be locked in and compared to throughout the year.
A forecast, on the other hand, should change often. Start with the plan as "Version 1" and use the forecasting team's skills and tools to bring those expectations as close to reality as possible. Notably, a forecast should not change constantly, otherwise nobody is held accountable for performance. Forecasting teams should work with leadership to set a cadence of updated forecasts, typically monthly.
With each cycle comes changes to the predicted outcomes that could not have been seen as far ahead as when planning took place. For example, a hurricane sweeps across your state, closing your operations for several days, the Fed changed rate twice as much as expected with no signs of slowing down, or a virus spreads across the globe in a matter of months (who could have predicted that?).
Forecasts can use these monthly cycles to modify their assumptions, historical input, and even their methodologies to improve the accuracy of their prediction. Plans help us to aim high early on, and keep us honest about where we wanted to be. Businesses and forecasting teams will stand out by using both in tandem, walking that tightrope balance of art and science, and using the difference or "Gap" to solution for growth.
Why Will My Forecast Always Be Wrong?
Finally, you might be here today because I accused your predictions of being wrong. The second part of the title is also incredibly important in setting expectations for yourself as a forecaster. It's okay to be wrong. Instead, the question you should be asking yourself is "how wrong was I?"
As forecasters, we'll go to bed bruised and bloodied every night if we beat ourselves up over every wrong prediction. The nature of our work is incredibly volatile, and others' expectations of our abilities are often higher than they should be. Let's instead judge our forecast on a scale, which will give us some room for both comfort and learning.
I use the following rule of thumb to quickly scan a forecast performance report to assess the accuracy of the reporting and performance of the business. In this rule, I refer to "forecast variance" which simply means:
Actual KPI at the end of the period
-------------------------------------------------- - 1
The forecasted version of that same figure
If your forecast variance is ± 0-3%, you nailed it. Congrats, that's about as close as most businesses can expect for many KPIs. Take the win and move onto the next KPI.
If your forecast variance is ± 3-9%, you missed by a bit. Don't worry yet, but if this becomes a pattern, revisit your forecasting methodologies and double check your assumptions. Try tweaking these levers until you reduce the variance over several cycles. You may not be able to find the root cause of this small variance, but all the better if you can before bringing it to other teams.
If your forecast variance is ± 10-15%, you missed by a lot. Make sure you dig into why this happened. Maybe there was an unexpected variable, such as a change in operations or supply chain issues. Definitely revisit the historical data, assumptions, and methodology in the next forecast cycle.
If your forecast variance is ± >16%, there's probably an anomaly. These are indicators that either something dramatic occurred in the core function of the business that drives that KPI, or that you made a mathematical error in your reporting. Prioritize these variances to investigate for errors, owning up to any calculation mistakes to stakeholders and leaders as soon as possible to quell worries. Revisit this when the next forecast comes around, and look for similar errors throughout the remaining forecast report. Fool me once and all that.
Scan your relevant forecast document and make note of anything you may need to discuss with stakeholders and leadership. Often leaders use these reports to drive day-to-day conversations and questions about what to prioritize and improve.
I'm sure there are some exceptions to this, so don't overreact if you have a KPI where this doesn't fit. Also, the more granular your KPI is, the more likely your variance to forecast is larger than the scale I provided above. Daily metrics may swing a lot more than weekly, and annual metrics take take shape over the course of the year, likely not fluctuating as much. Adjust this scale for your personal version of this rule so that it makes sense in context.
As a forecaster you play a vital role in making predictions and setting expectations across your business. Whether you run this on your own along with dozens of other tasks at a startup, or are a part of a large team of sophisticated statisticians whose sole job is forecasting, set yourself up for success by clearly communicating and thoughtfully assessing both forecasts and plans.
Ryan is a long-time data analysis and strategic management expert from his time supporting Disney Parks and Resorts, McDonald's Owner-Operators, high-speed rail, cruise lines, observation towers, family entertainment centers and more. He is based in Seattle, WA and specializes in pricing, forecasting, and revenue management for food and beverage operations, but loves the broader travel and leisure industry. In his free time, you might catch him as a Game Master in a tabletop game, cooking up a comforting meal, playing video games with friends, or reading a book in the park.
