Ste. Michelle Wine Estates and AllRecipes.com pair wines with AI.
by Dave Hanson
Artificial Intelligence is opening new doors to deliver shopper insights and turn them into real-time shopper marketing solutions. Just the other week, Ste. Michelle Wine Estates and AllRecipes.com launched into a fascinating new partnership. The companies will be working together using artificial intelligence to instantly deliver appropriate wine pairings alongside the recipes for users on Allrecipes.com.
The partnership is an intriguing convergence of real-time shopper insights, marketing and technology in a scenario were the output actually delivers real value to customers. The program classifies recipes to the appropriate pairings through machine learning techniques and then displays the suggestion next to the recipe. If you’ve ever been stumped searching for the right vintage to pair with an evening’s cuisine, well, problem solved.
What’s interesting and a bit different here is the building of AI recommendations based on recipe attributes. While most AI platforms make recommendations via buying patterns and browsing histories, this AI is identifying patterns in the recipe ingredients. It’s not reinventing the wheel entirely, but it’s an interesting and less common implementation of pattern recognition, and it has been applied in a pretty clever way here in the role of sommelier.
The program gets even deeper and more robust as we dive deeper into it. According to an article published the other week by mediapost.com the program also provides real-time insights such as location data, store proximity, impressions by channel, trending products and trending recipes. Furthermore, the wine pairings are linked to logical retailers for the user, such as local grocery store websites or ecommerce partners through Allrecipes parent company Meridith Corp.’s proprietary Shopper Marketing platform.
It might be buried well behind the lead, but that is a tremendous treasure trove of Shopper Insights data being delivered alongside the real-time marketing output. Having access to this kind of big data could certainly open up a lot of other potential opportunities going forward.
From every angle, this appears to be a really cool and forward-thinking program with a ton of useful outputs – both in terms of data gathered and marketing outputs delivered. It will be interesting to see how this type of product-attribute-based AI algorithm is repackaged to create useful insight for “recommendation marketing” in other categories going forward.