Accurate forecasting is a severely underrated skill in modern business and one that you should be looking to improve if you want to maximize the value of your planning and analytics. But the only way to do this is to have a feedback loop where you can look back on the forecast methodologies that you used in the past and assess how they fared when coming up against the uncertainties of the real world.

It’s for this reason, that a forecast post-mortem can be such an eye-opening component of your FP&A journey.

What is a forecast post-mortem?

As the name suggests, a forecast post-mortem is when you look back on forecasts of the past and evaluate how close they were to the actual figures. Many businesses are familiar with the general idea here which includes calculating and interpreting the variances on each line item. However, this is merely scratching the surface of what a forecast post-mortem can be.

The real value comes in digging into those differences and unpacking the assumptions that were used in coming up with those numbers. This is because natural variances are always going to exist, we don’t have a crystal ball. The only thing we can actually control for is the assumptions and drivers we used to make the forecast itself.  And so if we can compare those to what actually happened, we can fine-tune how we think about our industry, our company, and the world at large.

Preparing for an effective forecast post-mortem

The first step happens at the time when you actually make the forecasts, and that is to document your assumptions and drivers clearly and thoroughly. IBM Planning Analytics / TM1 is a great help here because it can house details about cost and revenue drivers at various levels of granularity ­– giving you true transparency into how those numbers were calculated. But regardless of what system you use, taking the time to note down the key building blocks underlying each number means that you have a baseline to compare things to when you eventually get to the post-mortem itself.

It’s also worth making some qualitative notes about the thinking and strategy that underlies the forecasts because these also add a lot of colour to the situation. Don’t think that you’re going to remember this in a year’s time – because you won’t. Dedicate some time upfront to document the process and you’ll be in much better shape to get the value out of the exercise.

What should you be looking for during a forecast post-mortem?

Here are some important considerations for doing your own forecast post-mortem:

• Select the correct metric to measure forecast accuracy. Mean Absolute Percentage Error (MAPE) is generally accepted as the best method to measure forecast accuracy. However which level of granularity to calculate MAPE is also important. As always with forecasts, make sure your granularity is correct.

• Focus first on those variances that are the largest in absolute percentage terms, as these represent the biggest outliers. Dig into each and identify which assumptions were wrong, and try to unpack the causality behind them. On the other side of the coin, draw a line of materiality so that you’re only investigating things that will actually move the needle for the organization.

• Look for patterns across your variances (over time and within a singular reporting period) as this may point to a systemic forecasting error. For example, if you’re consistently over budget for your fixed costs, it’s worth speaking to your cost accountants to get to the root of what’s happening. Sometimes it can be as simple as a misunderstanding between team members, and that can solve a lot of other problems downstream.

• Don’t ignore instances where your forecasts were near-perfect either, because those can also contain a lot of information in terms of what you’re getting right. If you’ve stumbled onto a highly effective forecast technique for a specific item, see if you can apply it elsewhere to improve accuracy in other places.

• Get input from as many people as possible. Forecasts within large companies are complex and collaborative processes that require a lot of coordination to get right. As such, try to involve as many stakeholders as you can in the post-mortem so that everyone can learn from one another and increase the skill level across the organization.

These are just a few of the components of an effective forecast post-mortem and if you follow these principles, you’ll find that you can optimize your forecasting and make better business decisions as a result.

Apliqo builds tools and offer support to organizations that are running complex forecasting efforts and want to get the most out of them.  If you’d like to explore how we can work together, be sure to get in touch today, and let’s see how we can help.