For several months now, the Supply Chain has been hit hard by an unprecedented crisis. Firstly, the arrival of the Covid-19 in China led to breaks in the supply chain, quickly highlighting the dependence of the world economy on China and pointing the limits of globalized supply chains. Within weeks, the spread of the virus and the accompanying health protection measures then impacted each organization locally. In addition to having to adapt to very brutal variations, logistics organizations also had to organize themselves to ensure, when possible, the continuity of their activity, despite the organizational difficulties involved in confinement.

Data at the forefront

The dependence of supplies from an almost unique production area and the lack of visibility of its entire logistics chain are today pointed out by experts. Visibility and information sharing between partners and suppliers are key elements of risk management and business continuity. The digitalization of the Supply Chain at all levels, already at the heart of concerns before the crisis, now seems inevitable and urgent.

The dependence of supplies from an almost unique production area and the lack of visibility of its entire logistics chain are today pointed out by experts. Visibility and information sharing between partners and suppliers are key elements of risk management and business continuity. The digitalization of the Supply Chain at all levels, already at the heart of concerns before the crisis, now seems inevitable and urgent.

Among all the links in the chain, the last mile is recognized as being the most expensive, the most complex to operate and the most dependent on reliable and quality data. In the midst of a health crisis, data and the means to activate it are a rare commodity that must be exploited to the best.

The last mile tested by Covid-19

In the test of the crisis we are going through, the last mile and its actors undergo brutal and very different variations in activity from one organization to another. The quarantine and the closure of a large number of outlets have had a severe impact on the distribution models in place. In mid-March, while Mondial Relay and Relais Colis announced a stop of their activity, Amazon announced that it would need to recruit 100,000 additional employees to meet the explosion in demand. Since the beginning of the crisis, many players have reviewed their business model and have adapted to address their market differently.

Variation in volumes and geographic distribution of orders, reduction in available resources, organizational constraints of social distancing, prioritization of certain channels or services: manufacturers and logistics specialists who were able to ensure the continuity of their activity had to be agile and reactive.

However, many of them operate with the means at hand and the vast experience of their teams. Most are indeed little or not equipped with the capacity for analysis and optimization to test different models and ultimately make better choices. Data and AI can, in this context, prove to be a major asset in avoiding very costly errors, and trying to preserve performance during the crisis.

Data, AI and Machine Learning for more agile logistics

In times of crisis, when transporters and logistics specialists are faced with strong uncertainties regarding their activity volumes and the presence of their employees, it is easy to understand that the optimization tools they use on a daily basis can be useful to them.

But are they really? Or rather, under what conditions do they remain so?

Optimizing delivery costs requires knowing the cost structure and parameters that apply to routes. However, in the conditions of the health and economic crisis that we are going through, many parameters have changed: fuel prices, equipment costs for protective equipment, need for overtime, lack of road traffic, etc. We must be able to reflect these changes in the optimization so that it remains relevant.

Food specialist players delivering the food service industry began to distribute their products at home, facing new constraints specific to B2C delivery. Yesterday, delivering at home implied scrupulously respecting the announced time slots under penalty of failing to deliver. At a time of confinement and absenteeism, this objective seems derisive compared to that of minimizing resources. These new objectives and these new constraints must be able to be declared and modeled in the optimization tools, without which the latter would remain in vain.

Adapting the use of operational tools to the new context in order to react to it is one thing, being able to anticipate and make the best choices with full knowledge of the facts is another. Today, few logistics specialists are equipped with solutions to simulate different optimization scenarios for the last kilometer.

By mixing machine learning techniques (on short and recent histories) and optimization, it is for example possible to test new configurations and new organizational schemes in the warehouse in order to evaluate them and predict their repercussions before they are put into practice (for example: closure of a site and reallocation of resources, abrupt reductions or rotation of staff, anticipation of service provider needs, modification of the number of truck departure waves, simulation of sequenced schedules, variations on cut-off times, etc.).

Once the new target organization has been identified with the simulation tool, it should be possible to automatically implement it in the optimization tool used on a daily basis in order to pass on the strategic choices to operations.