Key steps to improve forecast accuracy

Perhaps one of the most common issues I come across in business, and a key contributor to a variety of associated problems, is poor forecast accuracy.
Poor forecast accuracy can affect a whole facet of key (critical) elements from cash through to resources let alone the obvious issues like inventory and customer deliveries.

Unfortunately, many businesses treat forecasts as goals i.e. my goal is to sell 10 units next month.

Goals are not forecasts.

True demand forecasts are based on data, mathematics and business knowledge. They are the likely outcome of what’s going to happen based on evidence. Forecasting should look at specific periods of time, include inputs from various factors, include an algorithm (something that takes the inputs and turns it into a forecast and includes a forecast error rate).

Unfortunately, goals often get mixed with forecasts and the results end up being somewhat unrealistic. There are all sorts of reasons for that, management needing to hit sales targets, a belief that stretch targets are motivational to name but two but the key thing is Forecasts are based around data not a wish list and getting it wrong as we said in our intro drives problems.

Forecasting is a science and there are a multitude of complex algorithms that can be employed to help drive the process but there are also some fundamentals of the process that can help deliver great forecasting. So what are these steps? They include:

· Business consensus around the forecast – it’s not just the managers choice, Good forecasting will take inputs from various sources and build a consensus amoungst the stakeholders through following a process.
· Which leads me onto: A well-established / Adhered to forecasting process
· A “learning” culture that analyses accuracy of previous forecasts with the intent of improving future results
· Takes into account key business metrics & performance
· Makes clear it’s assumptions and risks
· Have special processes for dealing with New Product Introduction
· Stakeholder Collaboration

That last point is crucial. Collaboration between key stakeholders in the business is essential to a well-oiled forecasting process. But it’s not just within the business that collaboration is essential it’s also vital that this includes dialogue with key customers and the supply chain.
Common forecasting issues

Most business will have some form of Forecast accuracy KPI. I’ve been in various situations reviewing this where clients have blamed their supply chain for getting things wrong (usually not delivering) but often that’s underpinned by over zealous planning i.e. one that you were never going to achieve anyway!

Usually, most issues stem from one of two things, the forecast processes uses incorrect (or not enough) “ingredients” and a lack of incorporating a forecast accuracy factor. Lacking this forecasting bias within the process can result in either over estimating or under but a historical understanding of past performance and incorporating it into your process can improve forecast results considerably.

In some businesses, the sheer volume of SKU’s and customer/product mix can be daunting and can present an analysis problem with planners often citing that they have too much data to review. Of course, the obvious step is to look to aggregate this demand.

So what typifies bad forecasting?

Perhaps the key thing for me is where businesses don’t learn. They consistently fail their plan but do nothing to affect the root cause. Accurate demand planning is almost always about feedback. Failing to listen and act means that you’ll be consistently stuck in a quagmire of under-performance.

Have some thoughts on demand forecasting? We’d love to hear in our comments section below.