Virgin Wines is betting that “taste” can be modeled. And doing that will make wine buying feel less like wandering an endless aisle and more like having a trusted merchant who already knows your palate. The company is partnering with Preferabli and its “sensorial AI” to break down wines into sensory attributes (acid, sweetness, tannin, body, flavor profile) and then match those traits to individual customer signals gathered from questionnaires and behavior (searches, purchases, ratings). (AI Magazine)
Why this matters (and why it’s not just another recommender)
Most recommendation engines in retail start with patterns: “people like you bought X.” Wine is trickier because preference is sensory, vocabulary is fuzzy, and “I liked that Rioja” doesn’t automatically translate to “I’ll like this Barolo.” Preferabli’s approach is to build a taste graph that maps product attributes to personal preferences and uses it to guide discovery across channels. Virgin’s first move is personalized email for WineBank members, and then the plan is to expand into web and the upcoming app. (AI Magazine)
That sequencing is smart. Email is where you can test relevance (opens/clicks), trust (“they get me”), and conversion lift without rebuilding the entire storefront on day one.
This is the “discovery” war I’ve been watching for years
In 2019, I wrote that category-discovery apps like Vivino were vulnerable precisely because the giants already owned the distribution rails, including search, maps, photos, shopping, and identity, which could all unify the experience at a massive scale.
Virgin Wines’ move is the counterpunch: own the relationship and the preference data so discovery happens inside their ecosystem—not rented from Google, Amazon, or anyone else. In a crowded digital marketplace, differentiation is increasingly “who knows me best?” not “who has the most SKUs.”
“Machine intelligence with human touch” is the right framing. But only if they mean it
Both Virgin and Preferabli are leaning into the idea that AI supports, not replaces, expertise. The real value is when a retailer can blend:
- Machine pattern recognition (scale, speed, personalization)
- Human judgment (context, story, edge cases, trust)
Preferabli’s CEO talks about a world “driven by individual preferences, using machine intelligence with human touch,” and Virgin’s digital leadership is explicitly positioning this as a confidence-building experience. (PR Newswire). This is exactly what we have suggested to clients when it comes to AI.
That “confidence” word is doing a lot of work here. If they nail it, Virgin reduces returns/refunds, increases repeat purchase, and grows LTV because customers feel understood.
I’ve seen this movie in telecom. Just a different product, same playbook
In my writing on AI in telecom, I called out personalization as one of the most immediate, measurable wins: using customer behavior and usage patterns to tailor offers, recommendations, and experiences.
Wine is simply a more emotional category—taste, occasion, identity—but the mechanics are similar:
- capture signals
- model intent/preferences
- personalize touchpoints
- reduce churn, raise ARPU/LTV
The watch-outs (where “hyper-personalisation” can backfire)
A few things Virgin Wines will need to get right:
- Transparency & control. If customers don’t understand why they’re seeing a wine, recommendations feel creepy or random.
- Bias and feedback loops. Over-personalization can trap people in a “same-but-different” bubble and reduce discovery—ironically hurting engagement over time.
- Data and ethics. I’ve written that AI-driven personalization is powerful, but organizations need to be cautious about bias/discrimination and be thoughtful about how AI is deployed.
- Reality of the product. Wine is vintage-variant and supply-constrained. The model can be brilliant, but inventory and substitution logic determine whether the customer experience is actually “delightful” or just “almost.”
The bigger signal: retail is becoming “preference-first”
This Virgin Wines story fits the broader shift: AI is moving from a marketing add-on to the operating layer of customer experience—especially in categories where choice overload kills conversion. Luxury retail is doing it with AI-personalised homepages; grocery is doing it with taste-based discovery; everyone’s chasing the same outcome: make shopping feel curated again. (Vogue)
If Virgin Wines can turn “I like bright whites with texture” into a consistently great set of picks—across email, web, and app they’ll be competing less on price/promos and more on relationship.
And that’s where loyalty actually lives.