In recent years, more and more companies in the areas of supply chain management and materials management have relied on automated planning processes. The automation of planning processes refers to the use of algorithms and artificial intelligence (AI) to automatically generate forecasts for future sales figures and subsequently plan material and resource requirements automatically.
For production and trading companies, the automation of planning processes is of crucial importance for various reasons. First of all, it enables a considerable increase in productivity. Automation reduces manual tasks, avoids errors and frees up time for strategic decisions. It is also a solution to the growing problem of the shortage of skilled workers due to demographic trends. Our society is ageing, making it increasingly difficult to find and retain qualified staff.
Acceptance of automation often fails for eight main reasons
Of course, user acceptance of automated planning processes can only be expected if the forecasting and scheduling system is able to deliver reasonable results. In practice, this is rarely the case unless you have worked intensively on optimizing the scheduling parameters and put the resilience of the setting through its paces using simulation.
But even where we have created these conditions in our consulting projects, we encounter reservations among users about implementing automated planning processes. We often come across these eight main reasons for the lack of acceptance:
- Lack of understanding of how automation works: Many employees do not understand how algorithms work and therefore have little trust in them.
- Fear of job losses: Employees fear that automation will make their jobs redundant.
- Fear of errors: Some employees are afraid of errors caused by automated systems that they cannot notice or correct.
- Uncertainty about technology: There is a general uncertainty about new technologies and changes.
- Need for control: Employees want to retain control and therefore prefer to manage forecasts and orders themselves.
- Skepticism towards “black box” decisions: Decisions are made in a “black box” – not comprehensible to employees.
- Lack of confidence in the accuracy of forecasts and scheduling proposals: Long-serving employees in particular are convinced that their intuition and experience often lead to better forecasts than algorithms.
- Lack of adequate training and support: Many employees feel overwhelmed by the new systems and do not feel sufficiently supported.
The “classic” measures for greater acceptance only help in part
In order to overcome employees’ reservations and get them on board, the “classic” measures are repeatedly cited. I have also had good experiences with a pragmatic approach. But first let’s take a look at the classic measures:
- Education and training: When employees understand how the systems work, confidence in the new technology often grows.
- Clear communication about the benefits of automation: If the advantages of automation are discussed openly, this sometimes helps a little further. Automation is more likely to be accepted if the employees concerned are overworked and automation improves jobs rather than replacing them.
- Tolerant approach to errors: A tolerant approach to operating and decision-making errors can also help to increase user acceptance and reduce fear of the new solution.
- Employee involvement: At least some of the employees can identify better with the new automated planning processes if they are actively involved in the design of the automated processes.
- Create transparency: If you can make it clear to users how a supposed “black box” decision is made, this transparency also promotes trust in and acceptance of the systems to a certain extent.
- Demonstration of the resilience of forecasts and scheduling proposals: If it can be shown on the basis of simulations of past values that the automated forecasts and scheduling proposals have not performed worse than the users, this considerably boosts acceptance among superiors who are enthusiastic about technology. However, many users, especially those with many years of experience (to avoid the term “experienced”) often feel attacked by this.
Picking up skeptical employees with automation buffers
My additional suggestion: Why not try an “automation buffer”?
In day-to-day business, users are regularly faced with the dilemma of having to plan materials in such a way that the required delivery readiness is achieved while keeping stock levels as low as possible. Excessive stock levels lead to stress, as does insufficient availability of a material. A lack of material usually causes more stress than excessive stocks. A lack of material has a greater impact on operational processes than too much stock, production cries out, sales sends out reminders and sometimes even the management gets in touch. When I ask users why they want to closely monitor planning proposals from the planning system and not implement them unseen, I usually hear the fear of a lack of material, less often the fear of too much stock.
In such cases, I propose a small deal to the users, which we initially only agree for a few pilot articles as a test. I am negotiating with you about an additional stock that we will grant the material. This stock is intended to ensure that there is sufficient readiness for delivery even in the event of incorrect material planning: depending on how the planning solution is designed, this “automation buffer” can be implemented in different ways. In the DISKOVER system, for example, you can define a minimum stock level that will never be undercut, even if DISKOVER thinks it can manage with a safety stock below the minimum stock level. Another option is to define a manual safety stock factor, which can be used to increase the safety stock determined by the system.
Creating more trust in automation step by step
My experience shows that if the users agree to this deal and we agree an automation buffer, then they will also give the automated replenishment of these items a chance. After a few weeks, I take another look at the planning situation of the items together with the planners. We then usually find that the automation buffer has never or almost never been attacked, so we can agree to turn the automation buffer down a little. After a few rounds of optimization, at least a large part of the automation buffer can be reduced in this way. An essential prerequisite for this is, of course, that the system’s scheduling suggestions are correct, i.e. that the scheduling is set correctly.
If the pilot application was successful, we rolled out the automation buffer to other articles, thereby continuously increasing the degree of automation. In this way, it also becomes clear where the automation of scheduling is not yet effective, where humans should therefore intervene and where scheduling rules need to be further refined.
Significant side effect: the adjustment of scheduling parameters is avoided
The twist with the automation buffer offers another significant advantage. Users no longer feel compelled to hide inventory securities in different places in the planning system. Even if users have little knowledge of the planning algorithms in the software, many quickly find out how to achieve a higher inventory level by changing opening horizons, using manual safety stocks, increased goods receipt processing times, lead times in the system and other tricks, regardless of the delivery readiness level set. This means that scheduling parameters that have been set correctly with a great deal of effort are lost again if it is not possible to ensure that the parameters are set correctly and reliably adhered to, as is the case with DISKOVER via scheduling rules.
The productivity of our companies depends on the automation of business processes and therefore also on the automation of planning processes. Given the demographic facts, there is no way around this. It is better to take employees with us on the road to automation today than to find ourselves without employees and without functioning processes tomorrow!
Image rights: © by iStock