Let’s just say I’ve built a lot of planning models.

And, while drivers exist in all of them, I’d like to talk today about expanding the usage of drivers in planning. Most FP&A models rely heavily on collecting user input, either from department managers, or from centralized finance staff entering numbers. And, while some items are easiest to forecast by an individual deciding on what the number should be, much of any forecast can be a) automated and/or b) calculated based on drivers.

(Note that my experience is primarily building models in TM1 (IBM Planning Analytics), but this discussion applies to any financial planning model, regardless of platform.)

What’s a Driver?

Instead of forecasting in dollars, forecasting drivers involves determining the factors that drive a particular expense. Understanding your businesses’ expenses, which expenses are fixed versus variable, and how your expenses move with different external forces will hopefully give you a more accurate and relevant forecast.

For example, forecasting travel costs can be done by drivers simply by inputting the number of trips per month, and having a central parameter on cost per trip. Note that driver-based models still require inputs, but in this case we are forecasting the number of trips, and we can do some analysis by region or department to determine the best price/trip. This allows us to discuss this expense item based on business activities – what kinds of trips will people be taking, and for what purpose – and the expense is derived from that.

Drivers can also be external factors like weather, holidays, and inflation. These can be integrated using statistical methods to contribute to forecasting units sold, forecasting consulting hours, or forecasting interest income.

What are some common drivers?

Your planning model probably already has a few driver-based items. The most common driver we build into every planning model is Headcount, which drives all salaries and salary related expenses such as raises, bonuses, and benefits. As the headcount increases or decreases, all the expenses associated with people also rise and fall.

Revenue forecasting, also called Demand Planning, is full of opportunities for driver-based planning. As a starting point, forecasting Units and Price separately, and having revenue calculate as Units * Price, gives you two drivers to work with and analyze. This enriches the model, as users such as executives, sales people, and department heads can see the underlying Units and Price assumptions that went into any revenue forecast.

Utilizing drivers also gives you more “levers” to pull when you are doing what-if analysis and want to measure the impact of different business decisions. We can continue to peel back the onion –Price can also be driver-based, based on an historical average plus price increase, for example. By entering a price increase, an analyst can immediately see the impact on the bottom line. This automatically makes the forecast more explainable, and understandable, as the assumptions (price increase) are published in the model alongside the results (gross margin forecast.)

Key Performance Indicators are another excellent source of business drivers.  One of the easiest places to integrate KPIs into your forecast is for Balance Sheet forecasting. Dual entry bookkeeping means the Balance Sheet can be mostly derived from the Income Statement forecast. For example, receivables generally track with credit sales, via an AR days KPI. By lining up KPIs such as AR Days, AP Days, Inventory Turnover with the various Balance Sheet accounts, you can input just a few KPIs and have your Balance Sheet forecasted, since the data needed for a Driver-based Balance Sheet has probably all been entered into the forecast for the Income Statement.

More Drivers!

When you’re reviewing your planning model and looking for more ways to make it driver-based, consider these two categories: strategic items and normal business expenses.

For strategic items, having a driver-based forecast generally means having more of the assumptions about key parts of the business build right into the model. You’ve seen drivers in revenue models such as price and units, but also price increases, renewal rates, and other key revenue assumptions. The same can be done for expense items that are highly visible, strategic, or simply financially significant. Stock compensation can be such an item for some firms, with drivers such as new hires, tenure, and grant size.

For normal business expenses, the idea is to remove any need to review/forecast items that are generally not strategic and don’t need human review. For expenses such as these, simple drivers such as 3-month averages, growth rates, % of revenue or another expense, and headcount can be used separately or together to shorten the list of accounts that need to be forecasted directly.

Another key consideration is whether the driver is a mechanical driver or a statistical driver. A mechanical driver is generally the simplest to implement – it’s a simple formula like Units*Price. This kind of driver is generally built directly into the planning model, for example, in TM1 using Rules or Turbo Integrator. A statistical driver may be doing a regression analysis of past sales, with a growth assumption. These are valuable drivers where time series is key, or when external factors like inflation or GDP are a part of the equation. Many systems, including IBM Planning Analytics Workspace (PAW), have built-in time series forecasting which is a great place to start with statistical drivers. Integrating Python or R is also a popular method, regardless of your planning system.

The easiest place to get started is probably to look at what drivers you already have in your model, and where else they can be used. Headcount, again, is an excellent example. Your planning model probably is already using Headcount to drive salary-based costs. There are many other expenses that align neatly with headcount, such as laptop expense, rent, and other items that correlate with how many people you employ.

Tips for your Driver Based Model

Here’s a couple quick tips when delving into making your planning model more driver-based:

• This is an area where it’s important to involve the business users heavily – drivers should be chosen based on the key factors the business uses to analyze their numbers
• Start with just a few items to get people familiar with the approach
• Models that are too heavily driver-based often end up abandoned, because of the need to overwrite calculated values with target values from management, another good reason to start slow
• If you anticipate frequent “overrides” then you will need to work this into your architecture so that it’s easy to identify when overrides have been used
• Easy rule of thumb: for actuals drivers and KPIs are calculated based on the financials; for the forecast, the drivers and KPIs are input and the resulting dollars are calculated

Consider updating your planning model with some of these driver suggestions. The forecast becomes more meaningful, and hopefully more accurate, when it is based on clear business drivers.

If you’d like to get in touch, https://cubewise.com/reachout/

If you’d like to read more content like this, please make sure we’re connected on LinkedIn and reach out with a DM!