Millesima: Predictive AI to Maximize CRM Revenue

On the occasion of NRF Europe 2025, Sandra Fernandes, CMO of Splio, had the pleasure of taking the stage alongside Benoît Gibier, Chief Digital Officer of Millesima, and Anne-Laure Chauveau, Digital & Data Manager at Tara Jarmon.

The purpose of this talk was to show that CRM is evolving, particularly with the arrival of artificial intelligence, which is gradually changing the reflexes of CRM managers.

Following the article on WhatsApp based on comments made by Anne-Laure Chauveau, Digital & Data Manager at Tara Jarmon, we offer a brief summary of our discussion with Benoît: six questions to discover the strategy implemented by Millesima around the integration of Predictive AI into the daily work of its CRM teams.

Could you introduce us to the Millesima brand?

B. Gibier: Millesima is a wine merchant, a pioneer in online wine sales to individuals. Originally, we only sold Bordeaux wines through catalogues. Little by little, the brand expanded to reach more than 16,000 product references distributed across 20 e-commerce sites and 8 languages to manage on a daily basis.

What are the specificities of the wine market?

B. Gibier: At Millesima, we operate at the heart of a historically very traditional wine market, wherehuman expertise and trust have always been paramount.

When we chose to become a pure player in wine, we knew that these codes wouldn't change overnight. We therefore had to offer a personalized online experience, without ever falling into standardized communication, to remind them of the exchanges they enjoyed with their local wine merchants in the neighborhood.

Our customers reject generic messages: they expect precise advice, tailored to their tastes, their cellar, and the vintages they are looking for.

Your modern DNA is clearly marked, since you opted for predictive AI 8-10 years ago! What were you looking for at the time?

B. Gibier: Millesima has always been a very data-driven brand in its approach. We had strong desire: to remain authentic while accelerating the brand's growth in France and internationally. For us, this development meant industrializing personalization while respecting with industry regulations.

That being said, we faced several challenges, one of which was significant: we had a small CRM team. While we initially only sold Bordeaux wines, it was quickly identified that communicating about 900 producers would prove more complicated, particularly in the complexity of segmentation to decide which product to send to which customer, and in limiting marketing pressure.

So, 8-10 years ago, you chose Tinyclues, a predictive AI platform that was acquired by Splio in March 2023. How is campaign management working for Millesima via Predictive AI now?

B. Gibier: We quickly saw the added value of predictive AI coupled with our CRM thanks to the three-month POC conducted at the beginning of our collaboration. Manually, we were limited to two emails on Tuesdays, two on Thursdays, and one on Saturdays. With the growing number of countries to manage and producers to promote, predictive AI quickly allowed us to increase the pace of communications (3 to 5 newsletters on Tuesdays, multiplied by 20 countries) and to industrialize our CRM strategy. The marketing pressure exerted on our database contacts is nevertheless no greater because the platform allows us to limit the pressure per week while also taking automatic scenarios into account.

CRM management proves to be surprisingly easy. We no longer ask ourselves before each sand the classic questions of manual segmentation: when did they last purchase, what did they buy, what are their favorite products... Predictive AI takes care of identifying in our database who are both the most receptive to the email content but also those with the strongest purchase potential for the product or products highlighted.

What results could you share with the audience regarding your use of Splio Predictive AI?

B. Gibier: The first results came as early as the tree-month POC. We saw monthly revenue increase by 30% when segmentation was performed by Splio Predictive AI compared to a month with manual segmentation.

Open rates were also excellent, with peaks averaging 50%, while the unsubscribe rate was very low. We also noticed that our customers' purchase frequency had almost tripled! 

What recommendations would you have to convince others to switch to predictive AI?

B. Gibier: Without a doubt, if this choice had to me made again, I would did it my eyes closed! As long as the data stored in the database is properly structured (contacts, orders, product catalogue), predictive AI can only increase the revenue attributed to CRM. It's a virtuous circle that allows us to send fewer emails but with more qualitative and personalized messages. The tool has also allowed us to expand our international reach, despite having a small CRM team.

CRM with or without predictive AI? The Splio teams have conducted the investigation to enable you to transform weak signals into strong actions.