top of page

Is Autonomous Retail crucial in time of COVID?

Most software providers will tell you: Yes! As most of them already turned to AI to provide a different offer to retailers, using buzz words such as autonomous, self-driving, control tower etc.. the premises of this "generation of software" is to provide smarter solutions, helping through recommendation, easier and/or better decision making. But does it work? And do you need it now?

Thinking about autonomous and self-driving, we usually think of cars. And we can do numerous parallels, especially as many algorithms used in self-driving vehicles are shared with self-driving supply-chain. Let's start with a couple of questions :

  • Does the technology exist? Yes.

  • Do most car makers work on a self-driving option? Yes.

  • Do they all provide a viable option yet? No.

  • Do you need one now? ....

Most machine learning algorithms have been available for years, some of them for many decades and usually available online for free (here or here). With the computing capabilities increasing rapidly, those technologies are now made available for general purposes at a relatively affordable cost.

ADAPTABILITY is a key when considering software.

Ultimately most retailers are not buying a technology, or even a solution, but rather a service. A service which is tailored to their need, capable of providing value quickly, now and later. The retail landscape is changing rapidly- and this change is considerably accelerating in 2020.

From January to June this year, most people in the world will experiment some form on confinement, forcing them to change their consuming habits, accelerating the digital purchase, forcing retailers to better collaborate with manufacturers.

As the retail business keeps evolving, retailers need to make sure their solutions do the same. Here are some key criteria when selecting an AI service:

  • DATA: What is the process to manage the data in and out. Can new datasets be added easily? What is the process to add a new data set? \

  • BUSINESS RULES: the software should be driven by a set of rules which are managed easily by the retailer and can be updated in real-time to as the business is evolving.

  • SKILLS; most AI solutions are trained on SKILLS. For example understanding sales. understanding consumer behavior, promotions, inventories etc.... When selecting a software, make sure to understand what skills are available, which new ones are in the road map and what is the process to add new skills.

  • TRAINING: as new datasets and skills are added, the patterns of your business are also evolving and the AI solutions need to be trained regularly (Yes, Machine learning algorithms... actually learn ;).

Ultimately the software must be [consistently] trained for your business: although the same "Math" might be used, the forecasts of a food retailer have very different patterns and characteristics than the ones of a hardware retailer, the replenishment constraints would be very different than a soft-line retailer.... etc.

AUTOMATING - Yes, but what and how?

Of course thinking about AI is thinking about a different form of automation. From providing smarter insight, to recommendation and even predictions, AI technology can be used in many areas. What can be automated using AI ? (In the retail business), here are some areas generally covered by AI providers:

  • Forecasting - From regular time series, to the inclusion of internal events (such past / future promotions) and external events, the AI powered forecasting engines require less manual maintenance (Such as adjusting past events) and provide further accuracy. Those models should be trained regularly

  • Replenishment / ordering - Intelligent fulfillment includes standard multi-channel logistics constraints, and also consider data such as capital investment, dynamic service levels target (Down to sub-category level, ), supply-volatility, and capacity constraints of your SCM.

  • Consumer Analytics: Understanding the consumer behavior based of large groups of data including POS and loyalty information, market data, competitor data, customer feed-backs, accounting information etc... to provide fast moving trends at the most granular level possible.

  • Promotions: capability to analyze and plan the best promotions, recommend which promotions apply to for which categories and automate the process of allocation against the vendor trade funds.

  • Assortment optimization: using data from the forecasting engine, consumer analytics, and store constraints, the Assortment engine is capable of curating micro assortments and determine accurate stock levels with relevant product and category adjacencies, effective pricing, and profitable promotions.

  • Workforce: Ideally powered by the same forecasting engine than replenishment and other application, the workforce management should take into account all requirements already identified in the supply-chain and merchandising to ensure a smooth workforce planning.

Among hundreds of applications providing AI capable services, the successful ones will be focusing on your business. Making sure the application skillset, the business rules engine and the training are continuously focused on your business. The role of the human in the supply-chain will evolve from executing tasks to training tools, setting up objectives and anticipate what the machine can.

Although a self driving car can provide tremendous value on a day-to-day basis, in the middle of the race it will require a pilot (and a good one) to win the prize.

141 views0 comments

Recent Posts

See All


Post: Blog2_Post
bottom of page