Determing the right level of decision-making automation


John Westerveld has a great point in his blog post The right supply chain management priorities during the downturn position you for future success. For companies that want to make improvements now to be in better competitive position when the market recovers, there are several decisions to be made about which processes and tools to improve.

It’s amazing to see how many large multi-billion dollar operations still find themselves caught in the vicious cycle of widely using Excel spreadsheets. Some of them, recognizing that Excel is not the right tool and was not developed to technologies-for-decision-makingbe used in place of ERP systems look for alternatives that will enable them to become mere spectators in the planning and execution process.

I wonder how much thought companies typically give on deciding the amount of automation they want in their decision making process and how much is appropriate for the types of decisions that will be impacted by the new system.

I came across this Map of Technologies for Decision-Making, used by Professors Dwight Gertz and Tom Davenport at Babson College in Massachusetts. The picture proposes 2 dimensions that companies can use to organize their thinking about decision-making automation: the chart shows complexity of work in one axes and level of collaboration required in the other. This map can be used to analyze whether an organization is using or selecting the right technology and the right process given the nature of the decision they expect to support.

I believe that even in cases where the conclusion is that the process can be fully automated, companies need to be careful when planning for radical switches from fully manual to fully automated, for two reasons. First, it may require considerable change management effort to make everyone comfortable with all the calculations running behind the scenes; in some cases, a phased approach allows for increased confidence and hence better support to the project and better tool adoption (if you are not careful everyone will go back to the spreadsheets!). Secondly, even when you do a good job in the planning stage you should still expect to have to accommodate changes to the tool design during implementation: you’ll most likely see things differently and identify improvements. In addition, you need to ensure internal knowledge and documentation of how the analytics work so that, over time, as the business environment changes you know where to go and which setting to tweak in order to keep your system running up to date. Ideally, you’ll build in a process that would sense and identify the need for process reviews.  And you need to make the reviews when alerted, otherwise you may experience a version of the subprime mortgage fiasco, if you ignore the signs.

In any case, deciding what and how to change is a highly collaborative and judgmental decision, so don’t expect to find a tool to automate that one!


Francini Ortiz joined Kinaxis as a business consultant in 2008. Prior to joining the company, Francini held several supply chain positions at companies such as IBM and Alcatel-Lucent, where she successfully led global implementation of business processes and systems based on best practices and continuous improvement initiatives. Francini has an MBA with emphasis in Entrepreneurship, a BS degree in Industrial Engineering, and is APICS CPIM certified.

More blog posts by Francini Ortiz

Leave a Reply