In an era where leveraging data and artificial intelligence (AI) is crucial for business innovation and development, Manutan, the European leader in e-commerce specializing in distributing equipment and supplies to businesses and communities, and Factonics, a startup creator of the first 'AI Business Platform,' unveil the success of their collaboration

‍

In the exponentially growing e-commerce market, players continue to expand their product and service ranges using data collected through new technologies. The use of AI is becoming more widespread to enhance operational efficiency and customer satisfaction. With a product offering of over 630,000 items, Manutan stands as a major player in B2B e-commerce. Originating as the first French company to sell industrial equipment through catalogs, the Group evolved in tandem with the emergence of the internet and e-commerce, launching its first online stores as early as 2001.

In line with its rapid development, the volume of data shared by its suppliers has steadily increased, prompting Manutan to seek an efficient and flexible solution to optimize product classification on its e-commerce sites and support the expansion of its offerings. An AI solution was deployed to enhance product data classification...

‍

Complex and tedious, classifying unstructured and random product data provided by suppliers became time-consuming. Before encountering Factonics, Manutan had initially developed an in-house solution based on indexing tools to move away from manual processes, but the results in terms of properly classified products proved disappointing, not exceeding 10%.

"We reference an average of 130,000 new products per year, and our suppliers provide a substantial amount of data. We wanted to invest in an AI solution that would automatically classify products and optimize our internal resources. Factonics allowed us to address this business challenge and implement a solution tailored to our specific needs, thanks to their ready-to-use algorithms that we fine-tuned together," explains Guillaume Duval, Chief Data Officer at the Manutan Group."

Seven months after the initial meeting, Manutan conducted a first test of Factonics' proposed solution on a sample of products sent by its suppliers. This yielded an 80% success rate in correctly classifying products, an encouraging outcome that convinced the company to continue the project with the startup.

"The tool offered by Factonics has demonstrated its ability to relieve our teams in classifying supplier offers, tasks that were redundant for product managers and e-merchandisers at Manutan. Not only does their solution automate product classification, but it also enables more coherent product comparisons for our customers. Freed from these tasks, our employees can now focus on higher-value activities," comments Guillaume Duval.

... and to enhance the customer experience

From technical specifications to design, the way products are presented by suppliers varies, and the lack of uniformity in this information hindered not only the accurate classification of products but also their comparison. Factonics enabled sorting this information into major categories and sub-categories specific to the Manutan website.

Additionally, to address the lack of data for certain product categories, Factonics and Manutan initiated the collection of external data to feed the algorithm.

"We are already observing time and efficiency gains at Manutan, although the results are not yet consistent for all Manutan product categories; it promises long-term success. Currently, on average, we save 1 to 2 weeks per new offer referencing case, which translates to 280 days per year!" concludes Guillaume Duval.

The solution has evolved through the addition of algorithms and complementary alterations, and we plan to use even more new techniques to increase the algorithm's learning data and enable us to achieve a goal of 90% correctly classified products. With products better classified, customers can compare them more easily, especially their technical specifications despite data disparities."

‍

‍