It is with great confidence that I took a position as a procurement planner for Yoplait / General Mills early 2019. After all, I had several years of experience in the Supply Chain world. I started my career at Bel group for 4 years, joining as a production planner and then evolving in the implementation of an APS and becoming business owner for supply planning processes. I then joined the APS QAD Dynasys as a functional consultant and worked on their Procurement module, among others. Returning to the industry world in charge of supplying raw materials was supposed to be a piece of cake. 

The truth is, the reality was a bit different. My procurement plan was relatively easy to do, but it took me an excessive amount of time, because it was done in a huge Excel spreadsheet that took several minutes to open and even more to recalculate. To update it, I had to drag info from the ERP (stock levels, and an incredibly heavy file for BOMs) and retrieve the latest production plan sent by the planners in Excel. Only then could I launch my Excel-MRP. The worst part was that after doing this job, it was very frequent that an amended production plan was sent and I had to start over again. This led to frequent meetings with the production planners and demand planners in the hope of stabilizing plans, but the output was always the same: you can’t stabilize the demand. The customers are always right and our job was to move mountains to try and satisfy them.

A new approach to procurement planning seemed necessary. This is why I was so intrigued the first time I heard about Flowlity. Flowlity provides a whole new approach to mitigate volatility and stabilize plans. In addition,  the company is  building on the latest trends of the Supply Chain world and of Artificial Intelligence to provide its customers dynamically with the best plan at the lowest implementation and training effort.

Let’s try and understand the limitations of the most-spread approach, and what can be done differently.

 

How come MRP results are often very little used by the planners?

The traditional MRP approach to calculate the components’ requirements based on the sales forecast and a bill of material is correct in theory but has proven itself inefficient:

  • Complex and time-consuming settings of the BOMs and of the MRP parameters are a prerequisite to get the expected results. Often, the outputs are good just after the implementation of the MRP thanks to the expertise of a consultant or a key user, and get less satisfactory quickly as time goes by and the environment evolves. As a workaround, planners often resort to Excel files to calculate a simple BOM explosion better than in a wrongly-set tool, with all the risks of errors and inertia this implies.
  • The deterministic approach of the MRP is wrong by its very nature: it sets in stone “parameters” that are actually variables such as the lead time, the processing time, the scrap rate… Their variability has a major impact in the quantity and date of the consumption of components, and can’t be ignored as if the components’ consumption was a simple mathematical equation.
  • MRP-based approach creates volatility in the projections. Planners know for a fact that two sets of sales forecasts (on the finished products), even in a short time span, are likely to be very different from one another and will entail two drastically different plans on the components once the MRP is run. With this approach, the volatility of the sales forecast is directly transmitted to the components, when the planners need nothing but stability to make the right decisions.

 

What is Flowlity’s flow-based approach, and how can it help tackle those limitations?

In light of those observations, we offer a brand-new, flow-based approach inspired by the latest trends in Supply Chain Planning: we use advanced machine learning algorithms to generate a consumption forecast to provide you with reliability and stability with little implementation and training effort. What does it change?

  • Easy implementation and maintenance: forget about BOMs and MRP settings – all we need is to retrieve your consumption history (often easily extracted from your ERP), and we can start generating forecasts that get better and better thanks to our self-learning algorithms. No extensive project or training effort is required to start using the tool. Plus, variable settings such as the lead time are considered as an input for our calculation but are constantly assessed and challenged by our algorithms to reflect deviations between the parameter and real life.
  • Get a stable components plan: this approach mitigates volatility instead of cumulating it as a classical MRP is doing. Instead of immediately impacting the components’ plan to account for any change in the sales plan, we consider the long-term history (2 years in average) to calculate a trend and enrich it with the short-term events. The planners spend less time firefighting and starting everything over each time a sales forecast is updated and can focus on more value-added tasks.
  • Instead of leaving you with “one number” meant to be the truth, Flowlity recommends to keep your inventory level between a minimum and a maximum boundary to avoid stock-outs and overstocks. This intuitive approach makes the decision-making process easy for planners and is also an important driver of user-adoption.

 

How different is it from the DDMRP approach?

DDMRP approach also recognizes the limitations of the MRP and especially its complexity. DDMRP is about bringing simplicity back in the stock management process:

  • Thanks to a simple buffer system to mitigate the risks
  • Thanks to a simple and visual replenishment management system to give planners control over the stock levels

 

DDMRP has revolutionized the Supply Chain since its creation by Carol Ptak in 2011. Flowlity fully acknowledges the value of this approach. In fact, our philosophy is inspired by these principles (buffer positioning, consumption sensing, orders replenishment…) but wants to go even deeper by tackling two drawbacks of DDMRP:

  • DDMRP relies on static settings (like the coefficients based on volatility and lead time). The parameters need to be updated based on empirical knowledge of the field situation. In this case, how to make sure that the settings are correct? And how to update them fast enough when the volume is quickly changing? How can we really tailor those parameters for the entire portfolio?
  • The DDMRP doesn’t consider the optimization of the extended Supply Chain in order to stop the bullwhip effect. It is reductive to try to optimize stock levels by focusing only on one individual company and forgetting it is a link in a chain made of multiple suppliers and customers.

 

By getting inspired by DDMRP principles but dynamically adjusting our minimum and maximum stock recommendations and continuously improving our forecasts, we manage to get additional inventory reduction in comparison with the DDMRP for the same service level. The benefits are even higher when Flowlity is connected to other links of the chain by being the trusted third-party between the companies.

 

⎼ Written by Cécile Degouge, Customer Success Manager at Flowlity.