The wine service crisis hiding inside every restaurant's P&L
Why your floor staff isn't the problem — and what the smartest hospitality operators in Europe are doing in 2025 to unlock wine margin without depending on a sommelier they can't hire.
*Why your floor staff isn't the problem — and what the smartest operators are doing about it.* There's a conversation happening in every restaurant group, hotel F&B meeting, and hospitality trade show in Europe right now. It goes something like this: > "We need to do something about our wine sales." Then someone mentions training. Someone else mentions the sommelier vacancy they've been trying to fill for four months. A third person mentions the budget. The meeting ends. Nothing changes. Next quarter, the same conversation happens again. This isn't a people problem. It's a **structural** one — and it's costing the hospitality industry hundreds of millions in unrealised revenue every year. ## The numbers nobody wants to read Wine is, on paper, one of the most profitable categories on any restaurant menu. Gross margins of **65–75%** are standard. A well-executed wine programme can represent **30–35% of total F&B revenue**. In fine dining, that figure climbs higher. Yet most restaurants chronically underperform on wine — not because the list is wrong, but because **the conversation never happens**. The industry's own data is uncomfortable: hospitality staff turnover in Europe runs between **60–80% annually** in full-service restaurants (Hospitality Workforce Report, UKHospitality / HOTREC). Every departure takes accumulated product knowledge with it. Every new hire starts from zero. A sommelier earns **€40,000–€65,000 per year** in Western Europe — when you can find one. In Germany, the Netherlands, and Belgium, qualified sommeliers are functionally a scarce resource. Many independent restaurants haven't had one in years. The result: the most margin-rich category on the menu is being handled by the **least-trained** person in the room. ## What actually happens at Table 14 Picture a Tuesday evening. The restaurant is 80% full. Table 14 — four covers, celebrating something — is lingering over the menu. They ask the server about wine. The server, three months in, enthusiastic, genuinely trying, has absorbed the top five wines from a morning briefing six weeks ago. She knows the house red is "smooth" and the white is "fresh." She doesn't know the grape varieties. She doesn't know the region. She certainly doesn't know how to pair either with the dish the table just ordered. She suggests the house wine. They accept. The table spends **€48** on wine. Had the conversation been different — had someone with real knowledge been present — that same table, celebrating, leaning in, open to suggestion, might have spent **€90–120**. That difference, multiplied across 40 covers, five nights a week, 52 weeks a year, is not a rounding error. **It's a business model.** ## Why "more training" isn't the answer The instinct is understandable: train better, more often, more comprehensively. It doesn't work. Not at scale. Wine training programmes are time-intensive. **WSET Level 2** — the most basic credentialing that produces genuinely useful floor knowledge — takes months. By the time a new hire completes it, the odds they're still in the same role are, statistically, **below 50%**. And even fully trained staff carry knowledge imperfectly. Ask three servers to describe the same wine and you'll get three different answers — and probably one confidently wrong one. The deeper problem: **knowledge in the heads of individuals is not an asset.** It walks out the door. It calls in sick. It goes on maternity leave. It resigns on a Thursday to take a job in Amsterdam. Sustainable wine service can't be built on human memory alone. ## The shift the best operators are making The restaurants and hotel groups that are genuinely moving the needle on wine revenue in 2025 are doing something structurally different. They're treating wine knowledge as **infrastructure** — not personnel. That means making accurate, compelling, pairable wine information available **at the point of conversation**, regardless of who is having that conversation. It means freeing the floor team from the impossible expectation of memorising an evolving wine list in its entirety. It means giving guests — many of whom would rather explore quietly than ask — a way to engage with wine on their own terms. In practice, this takes different forms: QR-linked pairing guides, digital wine menus with producer context, and increasingly, **AI-powered wine tools** embedded directly into the dining experience — available in any language, at any hour, without a staffing dependency. This isn't about replacing the sommelier. **It's about making sommelier-level knowledge available when the sommelier isn't there** — which, for most restaurants, is most of the time. ## A note from the cellar At Pinhal da Torre, we work closely with HORECA partners across Europe. We hear this conversation in every market — from Amsterdam to Antwerp, from Hamburg to Lisbon. We've started piloting something we call **Pipa** — an AI sommelier built specifically for restaurant environments, trained on wine knowledge to WSET Level 4 depth, available in multiple languages, and designed to work **alongside** the floor team rather than replace it. The early results from our restaurant pilots are interesting. The conversations change. The upsell happens more naturally. The guest feels **informed** rather than sold to. We're still learning. But the structural problem the hospitality industry has with wine service isn't going away — and "hire more sommeliers" was never a realistic answer. The operators who figure this out first will have a meaningful advantage. If you're thinking about this problem in your operation, we'd like to talk. [www.pipa.wine](https://www.pipa.wine)