Forecast techniques you can use

Forecasting is critical for FP&A professionals to master if they want to stand out in their careers.

It’s a targeted way we add unique value to a company on a monthly basis.

Unfortunately, many FP&A managers don’t know how or when to apply certain forecast methodologies to get the best benefit.

Here’s a quick guide on how to think about applying different forecast techniques:

Start with a trend pick

A solid first step for a brand new forecast is to trend pick all of your metrics.

To do this, you need to evaluate historical data and simply continue the trend forward.

Have your sales been growing by 5% per year? Take your current sales level and multiply it by 1.05 for next year’s forecast.

Using a trend pick to forecast makes sense for situations like:

• Long term forecast assumptions that don’t require precision (pro forma)

• Expense accounts with a high volume of transactions (employee meals or office supplies)

• KPIs that are steady over time and don’t have a strong connection to other metrics (net promoter score)

Trend + Seasonality

You’ll quickly notice that a trend pick will fall short when there are months/weeks/days that tend to always be higher/lower than others.

For example, your sales team might have an annual bonus cycle. And they increase sales in December to max out their bonus.

Look if there is a clear monthly trend by dividing December sales by the average of all months in that year. If you see December is 150% of the annual average sales, then you can safely assume this trend will continue as long as the bonus program doesn’t change.

Best practice is to evaluate each of your metrics for seasonality when performing a monthly forecast.

Proportion of another factor

This is my most used forecast technique.

Finding a metric that you can anchor to allows you to predict a number of other metrics with ease.

One common way to do this is forecasting variable expenses. By definition, these expenses should move as a proportion of sales.

Here’s a few other common ways I use this forecast technique:

• Employee-related expenses that scale with headcount

• Daily sales that scale with the number of salespeople working that day

• Customer retention that scales inversely with price increases

• Call center staff that scale with call volumes

• Product theft that scales inversely with unemployment rates (this one works!)

Management input

Is your head of marketing excited about a tactic she is implementing to increase sales?

If so, consider using your trend’+’seasonal forecast as the baseline assumption then increasing your customer conversion rate based on the implementation of the tactic.

How much should you increase the conversion rate?

It depends on what the head of marketing committed to.

This technique is the only one on the list that doesn’t depend on historical data to make a projection. Meaning it’s often the least reliable, so use with caution.

AI & Machine Learning

It’s the talk of the town recently, but our Data Scientist friends have been using AI for years.

This one is dependent on your access to a Data Scientist. It also works best when you have a high volume of transactions that are difficult to analyze due to many characteristics.

Using machine learning can uncover patterns in your data that you have never seen.

You may learn that certain demographics of customers who buy particular products end up calling customer support more often. Therefore, you can forecast your future call center support needed based on your live sales data.

In Summary:

When forecasting techniques are applied correctly, you can systematize your monthly forecast.

Start simple with a trend forecast before you add seasonality, proportion of another factor, management input, and finally machine learning.

How do you plan to take action on this?

See you next Saturday.

P.S. If you are finding this newsletter helpful, then consider checking out my guide to become a world-class FP&A guru. The guide will teach you everything you need to know to accelerate your FP&A career.