How to build a driver-based budget

Driver-based budgeting, sometimes known as DBB, uses measurable activities from the business to build a financial budget. It can be a more realistic way of building a corporate budget because it highlights relationships between operational results and activities and the organization's finances.

There's a lot that impacts the business, so the important work is finding the right drivers to include in your model.

To identify your drivers, ask yourself these questions:

• What are the big goals that I'm trying to achieve?
  
• And what are the most direct activities that impact those goals?
  
• What are the relationships between these activities?
  

Getting to the right drivers usually takes some research. The best way to do this is to dig deep into each area of the business to understand how each function operates and impacts the overall business.

The most common mistake that people make in creating a driver-based budget is that they try to derive these values on their own or by only talking to the most senior-level executives. When they do this, the forecast becomes too generic to be accurate.

If you're able to marry the top-level view of the function with the granular details, you'll have a much better understanding of how each part of the business works and the different activities that impact its spending. For example, instead of just talking to the Head of Sales about your revenue forecast, talk with the AEs and SEs who would be able to provide you with more details on how the business is reacting to the current market environment. They would also have better information on what is affecting the business.

Similarly, if you're trying to improve your cloud spend forecast, then start with the head of product, who will help paint the vision for the organizational data structure, but also make sure that you talk to the specific engineers who manage the servers. The latter will be able to describe the specific activities that will cause spend to increase or decrease.

When you talk to these folks, don't make assumptions and ask a lot of questions. People are afraid of sounding dumb, so they don't dig in as deeply as they should. Verifying what might seem obvious can sometimes shine a light on counterintuitive parts of the process.

The next most common mistake that I see finance teams make is that they try to incorporate ALL of the detail into their model. When they do this, it causes a few problems.

1. You end up overfitting your model. Overfitting means that you tune your model to the exact past performance that you saw in the business, but since every quarter is unique, it ends up creating an inaccurate forecast of what will happen in the future.
   
   
2. The second issue is that adding too much detail to the model makes it difficult to maintain. The model becomes impossible for anyone else to edit, which ends up causing problems if you leave and makes it less usable for the rest of the organization.
   
   
3. Finally, including too many drivers usually creates an inaccurate forecast. Every driver ends up having some element of human assumption in it. Humans aren't great at predicting the future, so if we add too many of these, then you can find yourself really far off target.

Instead, focus on going deep to understand the full process, figure out which parts of the processes impact the numbers the most, and use those drivers in the model that you create for the business.

The next step after incorporating all the drivers and the business logic in the more is to verify that the model is working as expected. One of the easiest ways to do that is to use the inputs from say two quarters ago to forecast the previous quarter and compare it against actual.

1. Update all the drivers as the world looked two quarters ago
   
   
2. Set up the model to forecast the last quarter
   
   
3. Compare it with the actual performance of the business
   
   

The predicted values from the model would probably not match exactly to the actuals. The key now is to figure out why. Were these one-time events that are not expected to happen in the future or are you missing a key driver of the business?

Driver-based modeling is a really valuable way to create a budget or plan that reflects your specific organization. To do this effectively, it's important to understand how the drivers impact the business. This usually requires additional business context from the people who are involved in those activities.

The process of driver-based budgeting also requires that you find the balance between too much and too little detail. Too little will create an inaccurate forecast, but too much can cause overfitting and make the model difficult to maintain.