What Do Online Wine & Spirits Customers Reveal About Their Budget?
We analysed 912 real customer AI conversations across four continents to find out how customers talk about money when given a natural language interface.
| Sessions analysed | Continents | Languages detected | Highest single product click |
|---|---|---|---|
| 912 | 4 | 7+ | $5,022 |
Key Findings
- Customers volunteer budgets unprompted on every continent.
Across 912 sessions, customers stated spending limits without being asked. No pop-up. No filter. No prompt. Just natural conversation with something worth talking to. - Budget disclosure rates vary significantly by region.
European customers led with a 26.8% budget disclosure rate, more than 1 in 4 sessions. Asian customers disclosed at 10.7%, North American customers at 6.8%. The interface was identical across all stores. - Customers share context alongside their budget, not just a number.
Budget statements came wrapped in occasion, relationship and preference. “Wine to go with lamb, budget €85.” “Whisky gift for €50–70.” The number is always the last piece of information, not the first. - The multilingual reality is more diverse than any merchant expects.
Without any configuration, customers typed questions in their own language. In Australia: Spanish, Italian and Bengali, all in the same store, on the same day. - Operational questions are universal and consistently underestimated.
In every region, a significant proportion of conversations were not about wine or spirits at all. Customers treat the AI as a member of staff, not a specialist tool. - Gift intent drives the highest-value sessions everywhere.
From Chinese New Year whisky in Asia to anniversary wine in Europe, gift buyers share the richest context and carry the highest budgets. - Expert customers expect expert language across every region.
“A chardonnay that is not buttery.” “I like Guigal’s Côte du Rhône, something similar under €20.” These queries are impossible in a search bar. They are natural in a conversation.
A search bar captures “whisky under €50”. A conversation captures “whisky as a gift for a good friend I haven’t seen in years, it’s Chinese New Year, budget around $90.” Only one of those makes a great recommendation.
What Customers Actually Said
Verbatim customer messages from real sessions. Unedited.
- “money is not a problem” Europe, open budget, €150 interpreted by the AI
- “Wine to go with lamb budget €85” Europe, food pairing with stated budget
- “whiskey gift for 50 – 70 euro” Europe, gift intent with budget range
- “trouves moi un vin pour un barbecue” Europe, unsolicited French-language query, €100 budget
- “Wine to go with goan prawn curry tonight” Europe, hyper-specific occasion, €40
- “whisky as a gift for a good friend that havent met in a long time and its chinese new year. budget at $90” Asia, gift, occasion, relationship and budget in one message
- “I am looking at single malt about $200” Asia, premium product intent
- “budget is under $20 per bottle, it will be paired with chicken” Asia, food pairing with budget constraint
- “I’m having salmon tonight. What should I drink under $30” Australia, occasion, pairing and budget the click that converted
- “I love Tasmanian Pinot and eden valley Cabernet, have you got a pick that matches that mix” Australia, expert-level comparative query
- “Quiero un vino blanco para acompañar el salmón ahumado.” Australia, Spanish-language query, unsolicited
- “vino rosso, budget 40 Euro, da abbinare ai peperoni” Australia, Italian-language query with budget
- “Single malt whiskey under $60” North America, clear category and budget
The Numbers at a Glance
| Region | Sessions | Budget disclosure rate | Avg stated budget | Click-through rate |
|---|---|---|---|---|
| Europe | 183 | 26.8% | €35 | 8.2% |
| Asia | 328 | 10.7% | $71 | 11.8% |
| Australia | 327 | Early stage | — | 0.3% |
| North America | 74 | 6.8% | $20 | 14.7% |
Four Regions. Four Conversations.
The same AI Sommelier. The same interface. Four very different customer behaviours.
Europe 183 sessions
Nearly 1 in 3 European customers stated a budget unprompted, the highest disclosure rate of any region. Average stated budget: €35. Maximum: €150. French-language queries appeared organically in a non-French-speaking store and were handled automatically without any configuration. On one notable day, a single session produced 15 product clicks across 3 conversations.
Asia 328 sessions
Budget disclosure rate of 10.7%, with an average stated budget of $71, the highest average across all four regions. High-budget sessions were consistently tied to gifting occasions with rich personal context: Chinese New Year, anniversaries, reunions with old friends. The highest-value interaction recorded across the entire dataset occurred in this region: a single AI recommendation led to a customer clicking a product valued at $5,022.
Australia 327 sessions
Already showing strong engagement with a diverse, multilingual customer base. Spanish, Italian and Bengali queries appeared without any merchant configuration. Operational questions, stock availability across multiple store locations, delivery queries, discount codes, featured prominently, confirming the pattern seen in every other region: customers treat the AI as a member of staff, not a specialist tool.
North America 74 sessions
Budget disclosure rate of 6.8%, with budgets centred around $20–$30. Despite the lower budget range, this region achieved the highest click-through rate in the dataset at 14.7%, nearly 1 in 7 conversations ended with a customer clicking a product. North American customers showed a preference for direct, category-led queries with clear budget constraints.
The Language Finding
Across all stores, customers typed questions in their own language without being invited to do so and without merchants configuring anything.
| Language | Region | Example query |
|---|---|---|
| French | Europe | Trouves moi un vin pour un barbecue |
| French | Europe | un vin pour un cadeau d’anniversaire |
| Spanish | Australia | Quiero un vino blanco para acompañar el salmón ahumado |
| Italian | Australia | vino rosso, budget 40 Euro, da abbinare ai peperoni |
| Bengali | Australia | Real query recorded, non-Latin script |
| German | Asia | Ich weiß nur Beef Stroganoff. Budget so um die €15 |
| Portuguese | Asia | Preciso presentar uma amiga, pode me surpreender |
For retailers serving international customers, whether through export shipping, tourism, or a multilingual local community, a multilingual AI is not a premium feature. It is the difference between making a sale and losing a customer who could not get an answer they understood.
Key Takeaways
- Budget disclosure is universal but the shape is regional.
Every region produced unprompted budget sharing. But European customers disclosed at nearly four times the rate of North American customers. A global AI platform needs to understand and respond to regional customer behaviour, not apply a one-size-fits-all strategy. - Context is more valuable than the budget itself.
“Money is not a problem” tells you more than “€150”. “Wine to go with goan prawn curry tonight” tells you more than “red wine under €40”. The AI captures both and the context is what makes the recommendation great. - Multilingual capability is a basic requirement, not a premium add-on.
Spanish in Australia. Italian in Australia. Bengali in Australia. French in Europe. German and Portuguese in Asia. None of it was configured. All of it was answered. Every unanswered query in the wrong language is a lost sale. - Operational questions are everywhere and they matter.
Store locations, stock availability, delivery windows, discount codes, these questions appeared in every region. If the AI cannot answer them, it fails the customer at the first moment of trust. - Gift intent drives high-value sessions on every continent.
From Chinese New Year in Asia to anniversaries in Europe, gift buyers carry higher budgets, share richer context, and are the customers most underserved by traditional e-commerce interfaces. - Click-through rates prove the conversations work.
An 8–15% click-through rate from conversation to product across three established regions shows that customers who talk to an AI Sommelier are ready to buy.
Dataset compiled from 912 real customer AI Sommelier conversations across Europe, Asia, Australia and North America. October 2025 – April 2026.