Retailers have access to a lot of different data and are constantly looking for ways to gather better insights around how their customers are behaving, how their suppliers are evolving and how their distributors are operating. 

Dataiku provides retailers with a powerful accelerator to capture this data and develop actionable insights, including descriptive analytics and advanced modelling to generate predictions, according to general manager of business solutions, Sophie Dionnet. 

“Today, we do this for over 60 customers that represent the full range of players in the retail space. This includes the smaller, more agile digital native businesses that are advanced in data analytics and use machine learning daily in their operations. On the other end of the spectrum, we also work with the more traditional brick-and-mortar players, who are transitioning to becoming more data-driven,” she told Retailbiz in a recent interview.  

Benefits of AI  

It’s important to clarify how AI can be used, according to Dionnet. AI starts with accelerating usage of data, which can be either structured data or specific data points, such as the inventory a retailer sells, or unstructured data such as the data contained in documents or customer feedback. 

“AI is used to transform data into tangible applications, which can range from understanding and predicting sales and customer trends to monitoring and improving marketing inefficiency, managing supply chains or evaluating a supplier’s resilience,” she said.  

“Another area in which AI is used is in the analysis of customer feedback or social media sentiment to gain a better understanding of which products to stock, manage a retailer’s reputation or understand pain points to improve the customer experience. 

“AI is also used to help retailers understand product associations or customer hazards. This can translate to a number of situations that are key for retailers. For example, when a retailer is planning to launch a new clothing collection, they have to think about how much of the new product to stock, based on previous sales. The question of predicting stock is quite critical.  

“On the other side of the coin, there is always an expiry date to products — fashion can go out of style, or products simply have a tangible expiry date. Where AI can come is in creating specific packaging offers with the right associations to accelerate the sale of these specific products, or around understanding the optimal discount to encourage purchases. Additionally, AI can help identify the specific customers that will best react to the specific product or marketing message.” 

Demystifying AI to achieve business goals  

Dionnet believes advanced techniques such as AI are only useful if they deliver a return on investment by allowing retailers to make better business decisions.  

I would start by building an actionable list of use-cases leading to concrete outcomes, and delivering on these to create trust in the technology,” she said. 

“There is also a great amount of work to be done in helping analytics professionals and their business counterparts understand what AI is and what it is not. It’s not a magical tool, but it is able to help businesses generate better insights. However, these insights only have value if business leaders can act on them. 

“Another piece of advice would be not to automate too quickly. It’s important initially to keep the human in the loop as you build trust in the technology. There will be a moment when it’s time to launch fully automated campaigns, and you want to make sure that everyone understands why the technology is making X or Y decisions. So each stakeholder, everyone from the technical team, but also from the business side of things, needs to be on board before you can automate fully.” 

The evolution of AI  

Dataiku sees a difference in maturity between companies that are digital natives and those that are not. Digital natives tend to develop much faster because they are data-driven at their core. 

“On the brick-and-mortar side, they are more likely to be playing catch-up. While there are a few such businesses that are leading the pack, many are just starting to build their capabilities – which starts with building strong data strategies and first analytic applications, and exploring the potential of moving from descriptive to predictive,” Dionnet said. 

“Many of these players are still at the descriptive analytics stage — analytics that is used to understand an existing space versus actual prediction and automation. I think these companies will continueto develop, according to the maturity of their data structure. AI only comes from having the right type of data management foundations, which is something a little less native within the retail space. 

“When data science started in retail, it was focused on a number of areas, namely, marketing and supply chain. There are many other sectors, such as corporate efficiency and sustainability, that deserve muchbetter use of data. We are seeing these sectors start to come up more strongly on the analytics roadmap. 

“I would see a gradual, more advanced usage of AI and a diversification of use cases, notably towards supply chain and improved customer care.”