Data strategy and optimization models for more agility in the Supply Chain
Comparative views between adameo and Kardinal
Supply chains, more and more optimized, have become in recent years, real levers of business competitiveness. Despite this optimization, the dependence of supplies on centralized production areas and the lack of visibility and collaboration between the various links in the chain made the Supply Chain more vulnerable to exogenous shocks. The use of forecasting models that come as close as possible to real demand could then strengthen their performance by allowing them to adapt to the new consumption context. Indeed, in the test of the health and economic crisis that we are going through, the anticipation of short-term consumption has become more complex and no longer follows known patterns. If it remains at the heart of concerns, the anticipation of medium-term crisis exit scenarios also seems more than uncertain.
How then to manage one’s activity without having the capacity to foresee? In times of crisis, the Supply Chain must now be able to react as closely as possible to demand thanks to the digitalization of its processes and the use, at all levels, of advanced technologies allowing increased visibility and agility.
Distribution players are venturing into new models
The COVID-19 crisis has had a severe impact on the distribution models in place until then. The players in retail and mass consumption are subject to unprecedented tension: consumer concerns have encouraged large-scale “panic” purchases. Social distancing rules brutally made it impossible to consume outside of home, the usual share of which was passed on to distributors. Purchasing channels have also changed, with an explosion in the use of Click & Collect, home delivery and Drive, asking distributors to adapt quickly. The issue is important and the brands know it: if the user experience is there, consumption patterns could evolve definitively.
Many players have therefore adapted to maintain their activity. For example, to compensate for the closure of the markets, Rungis launched “Rungis delivered to your home” for residents of the capital, allowing producers to maintain part of their activity. For traders who only have a physical point of sale, initiatives are emerging and home deliveries are being set up, like in the chocolate industry (for whom Easter represents up to 30% of the annual sales). The government is also multiplying initiatives to encourage small traders to digitalize and move towards distance selling. New collaborations are also emerging, such as the association of Carrefour and Uber Eats to offer home delivery of essential products. Monoprix and Deliveroo aim to deliver 80 products “on average in less than 30 minutes, seven days a week”.
These inspiring examples, however, should not obscure the fact that for many players and many links in the supply chain, such a transformation in such a short time is too complex to operate. Absenteeism, social distancing, hygienic measures, all these new constraints are weighing on the activity of those involved in logistics today, and in particular on last-mile transport. In warehouses in particular, protection and adaptation measures are difficult to put in place.
Tools to optimize distribution at Covid-19 time
For some players, the drop in volumes inherent in the current context has made it possible to reduce the number of warehouse workers and has facilitated protective measures, including social distancing. Others, such as e-merchants, on the other hand, face the following double challenge:
- How to deal with a new, larger and unanticipated demand, with resources already under stress?
- How to organize field work, so that sorting of packages and loading is done while limiting the risks for employees?
The classic answer is to achieve a fixed geographic segmentation of the area to be delivered. Each truck is associated with a sector and will deliver the corresponding orders. This system greatly facilitates warehouse organization since the sorting and routing of packages to the loading bays is done according to the delivery sector. In practice, it often happens that the quantities to be delivered in a sector exceed the effective capacity of the vehicle. In this case, the supervisors rebalance, often manually and more or less urgently. This load balancing is carried out by the drivers among themselves, or by pickers who will transfer packages and pallets from one bay to another. Today, absenteeism and the necessary social distancing make these last-minute transfers almost impossible.
But developing a sectorization is very complex, because limiting load-balancing can involve a greater number of sectors, and therefore of vehicles and drivers. In addition, how can we ensure that the new sectors, potentially more numerous, will be more stable than the previous ones?
The question has deep mathematical implications and decision-makers today are ill-equipped to answer it. The use of innovative mathematical models could allow organizations to face new situations by making it possible to adapt the organization according to the objectives inherent in the new context.”Kardinal has developed a strategic sectoring model based on the use of Machine Learning, which jointly optimizes performance by minimizing resources, and the robustness of routes, meaning their stability before departure.”Jonathan Bouaziz, Co-founder & CEO of Kardinal.Used on a short and recent data history (representative of a few days of activity during the crisis), it makes it possible to resize the fleet to adapt to new constraints and to minimize handlings in the warehouse. If crisis exit scenarios with reliable forecasts are found, they can be integrated into the optimization model to accompany a gradual return to a “normal” activity at the end of the period.In logistics, data has been put aside for too long in favor of operational knowledge due to the lack of scenarios. Everyone agrees on the importance of collecting it, but its use is often not clear; and while the amount of data available is considerable, the quality is far from perfect. But a wave of major technological advances related to data (real-time geolocation, artificial intelligence, machine learning, etc.) now makes it possible to invent new logistics systems.
The importance of a data strategy for more agility in the Supply Chain
The implementation of a data strategy is today a crucial step for companies looking to gain in proactivity and in adaptability.
“Contrary to what many people think, making better use of your data is not just for large mature players but is well accessible to everyone, including SMEs!” Albéric Piot, Innovation expert at adameo
Regardless of the size of the business, assessing data maturity and building a realistic and consistent roadmap are the first steps. The implementation of an architecture and an organization to better exploit data and be able to extract value from it, will make it possible to gain visibility on activities and performance thanks to the implementation of functional bricks.
The data strategy will of course generate direct gains for the company in question, and will also have a positive impact on its external partners when data sharing is established. The bullwhip effect (a small variation in demand which causes amplified disturbances going up all along the chain of actors) which is very visible at the moment on certain essential products, can be controlled and reduced thanks to the sharing of data between the different actors in the chain.
“The technologies that allow us to extend a data strategy to an entire ecosystem of players are now mature. The collaborative supply chain is becoming a reality.” Pierre Olive, Head of lab by adameo
These diverse and more or less expensive technologies (collaborative platform, blockchain to secure exchanges, etc.) bring agility and responsiveness to a set of interdependent players. Each player keeps control of their data and decides access rights, but the entire ecosystem can react in real time to market developments, thereby varying their orders, production, storage and transport. Certified data sharing also allows better management of potential disputes and certain associated functions help automate tasks with low added value (invoicing, etc.).
Take the example of transporting perishable products. A delivery is underway to several recipients, with an acceptable time limit for sales (the product must arrive with a minimum durability date). An unexpected rise in temperature (a malfunction of the cooling system for example) takes place during the transport phase. The quality of the product is impacted and the time to sell this product while respecting a certain quality is therefore reduced. Thanks to real-time visualization of transport operations coupled with cost control, the company can react immediately. Instead of continuing the transport order to the initial destination, it will be able to redirect the delivery to a closer point for example, or even stop the current delivery directly to trigger a new one (because it could have observed that it is less penalizing to relaunch an order than to go to a quality dispute).
Optimizing your use of data and being agile means in short being more flexible, better at controlling your cycles and above all adapting more quickly to different market segments and customers. The uncertainty surrounding the current economic situation and the recovery model makes it all the more necessary to have the tools to help make decisions quickly. How to manage the activity for the next few weeks? How to adapt to high inconsistency? What resources to mobilize? How to optimize relationships with providers?
The list of questions that prevent business executives from sleeping at this time is long, but complex mathematical models and data enhancement technologies exist to help and support them in their choices for tomorrow!