Increasing wine sales by 250% with food pairings

We needed a retail partner willing to host our product in-store, but contacting them directly went nowhere. So we turned to our small winery customers already selling through retail to find a backdoor in. I interviewed the owners of Domaine Dardagny, a small winery selling Swiss wines at Corti Brothers, and heard the following pain points:

I hypothesized that:

• If we help shoppers find the right wine through pairings, they’ll buy more wine;
• driving sales for retailers;
• and in turn, we can provide wineries with the sales and shopper data they need to stay on shelves.

To validate this, I designed a two-week pilot to answer two key questions:

In previous PairAnything products, shoppers had to search for pairings. However, a blank search bar created too much friction in a grocery retail environment, where shoppers want to get in and out quickly.

To reduce cognitive load and drive users to the ‘aha’ moment faster, I replaced the search bar with tappable cards featuring images of Domaine Dardagny’s wines. Instead of thinking of a query and typing, shoppers could simply tap a card to see a pairing.

Before the redesign: The original PairAnything product only had a mobile view and required shoppers to type and search to get a pairing.

After redesign for kiosk: Shoppers are presented with 4 wine cards and can tap to see a pairing.

The card clicks revealed behavioral data about what actually caught shoppers’ attention, allowing us to report insights back to Domaine Dardagny that they wouldn’t have otherwise had. We learned that shoppers preferred white wines over reds and were drawn to tall, narrow bottles. In contrast, wines with softer curves and off-white labels received the fewest clicks, suggesting they were less visually appealing to shoppers.

Initially, shoppers could save pairings by creating a PairAnything account. However, through user research, we learned that no one wanted to make an account on a public kiosk, and if they wanted to save a pairing for later, they would likely use their phones to take a photo of the kiosk.

I changed the “Save Pairing” function into a “Send me more” button, where shoppers enter their phone number to receive more pairings via text. I wanted to test: Did shoppers want our pairing recommendations; so much so that they wanted even more?

15% of shoppers entered their phone number to receive pairings, confirming that they valued our recommendations. We shared insights with Domaine Dardagny about which wines shoppers were most curious about, along with a list of leads for targeted follow-up marketing.

To test PairAnything’s viability in retail, we needed to answer a fundamental question: Do pairing recommendations motivate shoppers to buy wine? Our CEO wanted to include all features to maximize impact, while our advisor warned that testing too many variables would muddy the results. I aligned the team around a focused, testable hypothesis: do pairings drive sales?

Faced with conflicting input from the CEO (who wanted all features) and mentors (who advised focus), I used breadboards and wireframes to explore options and narrowed the pilot to test one hypothesis: do pairings drive sales?

To isolate this variable, I trimmed the scope by cutting out non-pairing features and designed the test as a simple A/B comparison:

• 3 wines included pairing recommendations.
• 2 wines had no pairings, but prompted shoppers to enter their phone number to receive them by text.

This allowed us to measure the effect of pairings in a grocery retail setting without confounding variables. The outcome was clear: wines with pairings sold 3× more than those without.

One concern from the CEO was that if a shopper’s first tap landed on a wine without pairings, they may immediately walk away. To mitigate this, I ran an in-person user test by presenting the prototype on an iPad and tracking which wines users tapped first. The results showed that people naturally clicked on wines in the middle and along the edges.

I adjusted the design to make shoppers more likely to encounter a wine with pairings on their first tap:

• Wines with pairings were placed in positions #1, #3, and #5.
• Wines without pairings were placed in #2 and #4.

When shoppers click on a wine without pairings, they can read a tasting description and enter their phone number to receive the pairing later via text.

I designed each wine page to include a brief tasting description, so that every interaction provided value to users. Shoppers may enter their phone number to “unlock” pairings and receive them via text.