Email Twitter complexity Unlock Opinion Email Email Phone linkedin linkedin Download Down arrow

AI can help you optimize your stocks

15 Jan 2018
Most organizations have separate Sales and Operations Planning Departments. Naturally, these need to be aligned, so the company can work in an effective way. Until now, focus was placed on the alignment of individual plans – either manually through Excel or facilitated by more advanced tooling. In any case, manual tweaking and multiple meetings were always involved.

In the context of digitalization, however, the next step for Sales and Operations Planning is dawning. State-of-the-art technology enables companies to automate decisions that used to require entire meetings, and with the introduction of Artificial Intelligence (AI) into the business community, a new world of opportunities has opened up. One extremely interesting and relevant option for said departments is robust optimization for planning. In this blog, we will dive into the subject and shed some light on the opportunities it presents.


Sales planning: more information, easier choices

Forecasting the outcome of promotions has always been difficult. Setting up an advertisement is one thing, but analyzing and predicting its impact on your volumes is another. However, machine learning (ML) greatly facilitates the process of quantifying your impact – which, in turn, makes it easier to take decisions (i.e. using machine learning, it is possible to analyze a large number of variables with complex interaction buried in huge amounts of data with a high degree of noise. For example, data on promotions (including dates, discounts, and stores), media campaigns, and realized sales. Understanding the correlations among these variables would be extremely time consuming and difficult without machine learning). After all, it eliminates traditionally difficult choices that were solely made based on seasonal or weather-related trends.

 

Operations planning: aligning stocks with forecasting data

Besides being useful for sales and marketing purposes, machine learning for demand forecasting raises an interesting operations-related question: Why not use the information acquired to plan stocks and production? The neat trick is that the machine-learning demand forecast can provide both the predicted volumes and a level of certainty. The level of certainty is often ignored but can be of great value when used to plan your stocks or production. What if we use this level of certainty to determine the amount of safety stock needed to cover future demand fluctuations? This can be done through robust optimization, which uses the level of certainty as well as your desired service level to calculate the level of safety stock needed. (Up until now, safety stocks were set based on historic forecast deviations, discarding the information available in the demand forecast.)


Currently, it will require thorough data collection efforts and a considerable investment, as we are still dealing with complex systems. If you’re used to organize operations planning in Excel, you will have to up your game approximately three levels to achieve this goal. However, the development of machine learning tooling for demand forecasting is in full swing. Once larger companies embrace it, the creation of standard applications will be a matter of time. That is when it will get easier and more affordable to work with such a solution. But should an organization wait for this to happen, or is it wise to consider a switch now?


Robust optimization: what are the benefits?

Let’s take a moment to consider the benefits of a machine-learning demand forecast and robust optimization. First, through an ML demand forecast, you can gain better insights and eliminate many uncertainties that you have had to deal with traditionally. This will allow you to make more well-informed and fruitful choices. Second, through robust optimization, you will no longer have to work with forecasting deviations based on the past (looking at previous years to calculate an average for the future has proved to be inefficient more than once). ML demand forecasting creates a wealth of opportunities to realize a more efficient production planning, align stocks, and plan your margin. In short, quantification leads to cost efficiency – an objective that virtually every organization wishes to achieve.


Whether ML demand forecasting and robust optimization should be on a company’s priority list depends on a variety of organization-specific factors. However, one thing is certain: currently, the majority does not get the most out of the options available. Considering the rapid speed at which digitalization is developing, this should, at the very least, be food for thought.

Subscribe to our monthly newsletter

More articles