In past FP&A roles, I often used the slower days around the holidays to refine and improve my models. It’s a great time to step back and reflect, and ask the same question that I have posed to candidates over the years: What does good financial modeling look like in Excel? Below is my answer:
The model should:
… serve the objective.
Models are created for many reasons:
- to convince investors to invest
- to show the board your trajectory
- to accurately forecast the business
- to better understand the drivers of the business
- to motivate a team or the company
- to make a strategic or financial decision
Know who you’re building the model for, over what time horizon, and what outcome matters most. For example, a pitch to investors will look very different from a forecast used to run the business.
🎯Models exist to drive action—not to become works of art.
… incorporate historical data correctly (the baseline).
Of course, right? Nobody would disagree with this statement but the error that I have seen most often is either not incorporating or not understanding the historical data. The last thing you want to do in FP&A after a forecast cycle is to say, “we should have known this would happen.”
✅Understand the past before you predict the future.
… be driver-based.
The old mistake that people used to make is to say “we only need to get 1% of a market to be successful” without quantifying what that means in terms of salespeople, pipeline, pricing, units sold, etc. You should be able to back up any financial number with a couple levels of supporting operational math.
📊 Good models connect the big picture to the drivers underneath.
… balance simplicity and complexity.
Complexity should only be introduced in the model as a last resort. More complexity leads to model error and to increased difficulty in updating and maintaining a model. If a simple top-down model is producing a more accurate forecast than your bottom-up, assuming accuracy is your goal, then your model is likely too granular, or you don’t understand the baseline.
⚡ Keep it simple—focus on what’s forecastable.
… structured in a logical and easy to follow way.
The only way to understand what works and what doesn’t is to look at a bunch of models but a few points:
- Create a structure that is easy to follow without explanation, if possible
- Add tabs only when needed
- Group general assumptions in a single place
- Differentiate between inputs and outputs
- Etc.
🔑 Good structure makes models easy to navigate and understand.
… be easily auditable, with internal checks.
One of my pet peeves in a model is a hard coded number without a source. If you can link directly to a source, that makes it much easier to audit.
I also over index on internal checks that net to zero when the numbers are correct. Sometimes this adds to the messiness of the model, but I hide / group the checks, so that the model can be presented cleanly.
🔍 Make it easy to spot check the model for errors.
… go through proper reviews.
This is another statement that nobody would disagree with but does not always happen. A solid process looks like:
- One person builds the model.
- A second person checks for model error
- A manager reviews high-level logic, outputs, and explain-ability
Ideally, this is done before the CFO ever sees it.
🧩 Proper reviews catch errors before they become problems.
Sea Cloud Consulting 
Leave a comment