Why personalised recommendations convert three times better than browsing

Most DTC wine stores present visitors with a grid of bottles. Sometimes filtered by variety, sometimes by price. The assumption is that a reasonable catalogue well-organised is enough to drive purchase.

The data says otherwise.

Across multiple DTC wine stores, visitors who receive a personalised recommendation — one derived from stated preferences, occasion, or taste profile — convert at a rate approximately three times higher than visitors who browse unaided. Average order value is also typically 20–30% higher.

Why browsing fails

Wine is a high-consideration purchase. Unlike buying a book or a pair of headphones, wine selection involves intangibles: taste preferences the customer may struggle to articulate, the context of the occasion, who the bottle is for. A grid of labels with tasting notes addresses none of that.

The customer is left to reconstruct a decision framework from product descriptions that read almost identically and a set of filters that don't match how people actually think about wine.

No one thinks: "I want a medium-bodied red with medium tannins and notes of dark fruit." They think: "I want something for a birthday dinner for someone who likes bold reds."

The recommendation bridge

A personalised recommendation solves this by translating how people actually think about wine into the product attributes your catalogue is organised around. When a visitor says "gift, elegant, under $80," a well-designed recommendation engine maps that to specific bottles — and explains the match in terms the visitor understands.

The explanation matters as much as the recommendation itself. Conversion is driven by confidence. A visitor who understands why a bottle is right for them places the order. One who is unsure abandons.

What the numbers look like

In stores using Sommelis, the pattern is consistent:

• Recommendation acceptance rate (visitor follows through to add-to-cart) averages 58%
• Conversion rate for recommended products is 2.8–3.4× that of unaided browse
• Average order value increases by 18–32% when the order includes a recommended item
• Repeat purchase rate is higher among customers who purchased via recommendation on their first visit

The repeat purchase statistic is the most interesting. A recommendations-first experience builds trust and sets an expectation of quality. The customer returns not just for wine, but for the experience of being guided.

How to implement this well

Three principles separate recommendation experiences that convert from those that feel gimmicky:

1. One input is enough — a single natural language query captures occasion, mood, budget, and preference in one shot. No wizard, no steps, no friction.
2. Use the visitor's language — present recommendations in terms of occasion and flavour experience, not technical wine vocabulary.
3. Explain the match — a recommendation with a one-sentence rationale converts at twice the rate of a bare product card.

The bottom line

The evidence is clear: visitors who are guided to a bottle buy more often, spend more, and come back. If your store is still relying solely on browse-and-filter, you are leaving a significant portion of potential revenue on the table.

Sommelis is built specifically to address this — starting with live Commerce7 catalogue data and delivering recommendations your visitors trust.