In our last blog post we shared the outcome of an EyeOn questionnaire showing that a large group of companies envision the level of automation of the planning process to dramatically improve in the next 5 years. Companies are committed to taking steps towards increasing the automation of their planning processes. Analytics holdthe promise of two things:

  • increase the efficiency of tasks that once required substantial time and human effort
  • improve the quality of forecasts, plans and decisions through mining large amounts of data to discover new insights that were previously inaccessible.

Undeniably– analytics is changing planning processes – but quite some water has to pass under the bridge before companies will get to full no-touch planning.

From my work with renown clients I see considerable investments being made to enable the digital transformation in the planning domain. In the coming weeks we will share learnings on the digital journey using the EyeOn No – Touch framework.

Figure 1. No-touch planning framework

Figure 1. No-touch planning framework

Making the change – Clients are starting projects to discover the added value of analytics in planning, increasing staff with the right digital skill set, implementing new advanced planning systems and ways to collect, store and improve large data sets in data lakes.

Processes – Demand management, inventory management, supply planning and S&OP / IBP are all impacted by analytics and the usage of more internal and external data. In demand management we already experienced nice examples of improved promotion and new product forecasting using advanced models with external data. Sven Crone of the University of Lancaster demonstrates examples of applying Artificial Neural Networks that speak to the imagination. By applying a Forecast Value Add (FVA) metric it is easy to highlight where the system outperforms humans.

A company delivering electronical equipment decided to perform an assessment on the forecast ability of the demand for their 150.000 SKUs. It turned out that for 70 to 80% of the product portfolio statistical forecasting would outperform the forecast generated by the sales teams. As a result it has been decided to create statistical forecast team per region that delivers a statistical forecast to the sales teams. Those teams are responsible for the forecast of the remaining items using customer input into consideration.
 

Sensing demand changes and reacting to disruptions becomes much easier by having access to consumer reviews and sentiments. This allows for faster decision making in case of deviations and hence a reduced data latency.

Excellent data From several EyeOn events and projects it became clear that having excellent data is seen as a key enabler for further automation of planning processes. Although all the areas are needed, that does not necessarily mean that all areas will receive equal amount of focus from companies. When 23 business leaders were asked how they would spread their efforts over the next 2 years amongst the different building blocks, from the total of 100%, excellent data received the highest score with respondents allocating 28% of their efforts on average.

Figure 2. Effort spend in the different building blocks

The digital twin (a digital replica of the physical supply chain) requires the storage of large amounts of internal and external data.  A good example of using internal data for the benefit is the application of data engineering to keep planning parameters in sync with reality (using all relevant ERP transactions) leading to more reliable plans from Advanced Planning Systems (APS).

Application enabled In the last years, there has been a boom of new tools and platforms, at a high level some differentiating factors used by APS vendors are:

  • Industry focused solutions. By targeting specific areas, niche players have created differentiated solutions to compete with SAP broad approach. The OMP solution for the pharmaceutical industry or Quintiq in the process industry are well known examples.
  • State-of-the-art technologies at the core. Technological developments in areas like Machine Learning or Artificial Intelligence are embedded to make recommendations and speed up decision making processes. Companies like O9 and Aera are mentioned by MIT Sloan Management Review for having more AI and Machine Learning capabilities.
  • Easy connectivity with internal and external sources of information.The ability to easily integrate other sources of information in the data models allows for faster deployment and facilitates system architecture integration. Anaplan is often mentioned by Lora Cecere as one of the next generation APS applications.

The approaching SAP-APO deadline stimulates companies to reevaluate their planning system landscape and opens a window of opportunity for the new vendors.

High quality analytics – Applying Machine Learning and Artificial Intelligence in planning is propagated as the way forward. To transform processes to ‘no-touch’ decision making also has to be automated. Two different approaches are seen. First, fully automated decision making without human intervention. A nice example is the navigation system in your car adapting the route based on the latest traffic information, or one step further, self-driving cars. Advocates state that these will lead to less accidents because the car is faster in recognizing problems and deciding based on these changed circumstances. The second area is augmented decision making where the ‘machine’ makes a proposal but humans are still in charge of taking the final decisions. In the medical world the doctor gets a proposal but stays accountable for the final decision.

To what extent do we currently see (semi-)autonomous decisions in the S&OP process?  The level of decision automation is currently really low (1.1 on a 5 point scale) while respondents see possibilities to automate 50% of the decisions (Figure 3).

Figure 3. Current level of S&OP decision automation

Organizational readiness – Although no-touch seems to imply that humans are out of the equation the opposite is true; we still will need people. According to an article in Harvard Business Researchcompanies will need trainers (teaching the algorithms how to perform), explainers (elucidate the outcome of the machine to the non-expert users) and sustainers (assure that the machine are functioning properly, safely and responsible). New (groups of) functions are created with sexy names like analytics skunk works groups or data ninjas. Organizational challenges arise like how to attract and retain the correct staff, how to organize the analytics department and how to educate the remaining staff. A large part of the project budget at one of our clients goes into the creation of a digital academy to train the Operations and Supply Chain staff to be able to cope with the requirements of the digital era.

Decision focused culture The primary objective of planning is to make accurate decisions quickly and based on the best possible information. Ultimately, a company’s value is just the sum of the decisions it makes and executes. How to measure the quality of a decision? At the moment a decision is made the outcome is not known, and when the outcome is known, often a number of months later it is not possible to relate the outcome to the decision. In specific circumstances like in an operating theatre or in the cockpit of an airplane the quality of decisions is of lifesaving importance. Research has focuseon the concept of decision quality. Decision quality does (1) look at the ‘right’ answer’, and (2) if the most important parties are part of the decision process to achieve alignment and commitment to action.  The main elements that influence the making of the described best possible decision are:

  1. Process quality, the degree to which a process continuously ensures that the rules for information validation and for decision- making are sound. Procedural quality reflects how good the process is defined and executed.
  2. Information quality; is the degree to which correct and up to date information is available to support decision making.
  3. Behavioural quality; the degree to which countermeasures are applied to limit the effect of bias inflicted into the planning process.

Since it is difficult to actually measure the quality of a decision in a VUCA world,the decision quality framework defines the pre-conditions to make high quality decisions.

In the coming weeks we will share a blog post per improvement area providing more insights. November 14th we are hosting the EyeOn International Planning Inspiration Day in Amsterdam. The largest forecasting and planning event in Europe with more than 500 planning experts and over 30 content sessions! Please send me a message if you also want to join!

Freek Aertsen
Managing Partner at EyeOn
 

 

1 Wilson J., Daugherty, P.R. Collaborative Intelligence: Humans and AI are joining forces. Harvard Business Review. July – August 2018.

 

 

EyeOn is a niche boutique consultancy firm that works with clients designing, implementing and executing excellent forecasting and planning processes as a discriminating factor for success. In order to achieve this, we develop and share knowledge with demonstrable return on investment for our clients.