Properly applied, forecasts in sales controlling are essential for B2B companies. Nevertheless, they often fail. This article discusses why they fail and how you can use predictions successfully.

Next to reporting, forecasts are one of the main tasks in B2B controlling. Some B2B companies waste valuable resources on forecasting processes, which are not used as they should be.

If you have landed in this article, you have already realised the benefits of successfully deployed sales forecasts. Feel free to check out this article if you need to be made aware of its benefits.

Forecasting in sales controlling – What goes wrong? To say how forecasting processes can be used successfully in B2B sales and controlling, let’s first look at the places where things don’t go quite right.

1. Is the Forecasting Process adding value to the Company?

Automatize a bad process and you will have a bad process automatized. If a process does not work before automation, it will not work afterwards. A big mistake is to automate forecasting processes without first checking whether they are value-adding.

You can characterise a value-creating forecasting process by the fact that it is transparent for all parties involved, saves time and costs. In addition, data that flows into the process should be standardised, and the results should lead to to-dos that pursue a concrete benefit.

An example of a non-value-added forecast process in sales is documentation from the sales team with estimates of upcoming sales:

“How much will you sell to customer XY in 2023?”

These forecasts cost the sales team a lot of time and are also error-prone and usually not standardised. In turn, they are not used and go untouched in the archive. Maybe someone in controlling tries once a month to reconcile all this information. Cheers to our heroes!

2. Manual, Rule-based Forecasts.

Many companies invest time and human resources in manual, rule-based forecasts. However, the best manual prognosis is significantly worse than a predictive analytics system that uses machine learning algorithms with ERP data. Even if no forecast ever achieves 100% accuracy, it is desirable to optimise the precision of the forecasts and, at the same time, minimise the effort required to create them.

3. Lack of Knowledge about the Limits of AI in Sales.

When automating forecasting processes in B2B companies with the help of AI, there should also be a specific basic knowledge about what AI can and cannot do. Since a machine learning system learns from already existing data, the systems usually have a hard time with unpredictable events. In B2B sales, for example, this would be a significant customer deal that is newly closed. Such events can lead to a strong deviation between forecast and reality; therefore, you should consider them separately. However, hands up if anyone can predict the unpredictable. Nobody can, so do not expect a machine to can either.

4. Lack of Integration of Forecasts into the Daily Sales Routine

In many companies, the successful automation of forecasting processes fails due to the lack of use and poor integration into the daily sales routine. There are various reasons for this: lack of trust in the forecasts, insufficient sales team training or lack of processes that define concrete to-do’s based on the forecasts. One thing is sure: sales forecasts need to be followed by action to be helpful to the company.


Successful Automation of Forecasting Processes with AI

These sources of error in the automation of forecasting processes give rise to the following four recommendations for action for the successful automation of forecasting processes with AI in sales:

1. Machine learning and artificial intelligence are not an end in themselves. Companies should first develop a purpose – a “purpose”. Then technology can incredibly facilitate and improve the way to achieve it. That also applies to successfully automating forecasting processes with AI.

2. Make sure you choose the right technology for your business. You have a lot of data if you have thousands of customers and products. Here, investing in an AI-based predictive sales analytics system would be significantly preferable to manual analytics.

3. Learn about the capabilities and limitations of AI systems so you know what you’re getting into.

4. successful automation of a forecasting process always ends with concrete sales actions. Ensure end users are sufficiently trained and have clear action instructions based on the forecasts.


Further Read (german language):

Schäffer, U. (2022): “Ich muss ein klares Zielbild formulieren”. Hg. Springer Professional