AI Upselling in Hospitality: How to Increase Revenue per Table Without Extra Work

It is not only the busy evenings that determine whether your restaurant or hotel has a good month, but also how much you manage to sell per guest. An extra bottle of wine, a matching dessert or a package that is simply too good to refuse: these are the difference‑makers in your margins. With AI upselling in hospitality you present those suggestions consistently at the right moment, without your team having to spend extra time on them.

According to recent CBS figures on the hospitality sector, margins in the sector are structurally under pressure, while Koninklijke Horeca Nederland continues to cite staff shortages as the biggest bottleneck. Upselling intelligently without increasing pressure on staff is therefore one of the most direct gains AI offers in 2026.

In this article I explain how AI upselling works in practice for restaurants and hotels, which suggestions work and which do not, how to keep the tone so it does not feel like selling, and which pitfalls you should avoid.

What is AI upselling in hospitality?

AI upselling in hospitality means automatically offering relevant additional products or packages at the right moment in the guest journey. Think of a wine that matches the chosen main course, a dessert suggestion in the confirmation email, or a dinner‑and‑stay package with a hotel reservation. The AI uses context such as time, group size, preferences and history to make the suggestion personal rather than generic.

For the broader context of AI in hospitality, I recommend first reading the pillar on AI chatbots for restaurants.

Practical tip: a good upsell never feels like selling, but like advice from a colleague. Limit yourself to one relevant suggestion per moment rather than a list of options.

Where in the guest journey does upselling work best?

In practice we see four moments when AI upselling consistently delivers results, provided the suggestion remains short and relevant.

1. At the reservation confirmation

This is when the guest is most open to extras. A suggestion for a welcome drink or a vegetarian menu in the confirmation email or WhatsApp message converts above average in our experience.

2. On the day of the visit

A short reminder with a suitable dessert package or wine choice keeps your venue top of mind and triggers additional revenue more often than you might expect.

3. At the table via web chat or QR flow

Guests who are seated and ask a question about the wine list via your web chat immediately receive a suitable suggestion. Read more about the AI chatbot and how we implement this properly.

4. After the visit for the next time

A review request combined with an invitation for the seasonal menu or a private dining package turns existing guests into repeat visitors.

Which suggestions work and which do not?

Across the hundreds of implementations we have carried out, we see quite clearly which suggestions work and which do not.

• Do work: a matching wine with the main course, dessert choices in the confirmation, dinner‑plus‑overnight packages, children’s menus for family reservations.
• Do not work: aggressive discounts, suggestions that do not match the season or occasion, long lists of options or upgrades without clear price expectations.

Statista regularly publishes data on expected response times in customer service, and the pattern is clear: the more relevant and concise the suggestion, the greater the chance of acceptance.

How do you keep the tone human?

The biggest pitfall of AI upselling is that it can quickly feel like marketing. Three principles help keep the tone natural.

• One suggestion per moment, no more. A good suggestion does not need to be supported by a long list of alternatives.
• Use the guest’s name and refer to the chosen reservation or menu rather than using a generic greeting.
• Leave space for a friendly no, without the message feeling like sales pressure.

For the broader email approach see AI email handling and for voice and telephony see AI telephony.

The difference from traditional upsell campaigns

Traditional upsell campaigns are usually standardised suggestions that are the same for every guest. AI upselling differs in three ways.

1. Context‑aware: the suggestion takes into account time, group size, weather forecast and occupancy.
2. Personal: for returning guests the AI refers to previous preferences without it feeling like profiling.
3. Scalable: the same flow works for one location or a chain of fifty venues, with a local tone and menu per location.

A relevant background is the European AI Act, which the European Commission regularly publishes about. For upselling this mainly means transparency: the guest must know they are speaking with an AI.

Common mistakes with AI upselling

In practice we repeatedly encounter three mistakes, all of which can be prevented in advance.

• Too many suggestions at once, causing the guest to drop off or ignore the message.
• Suggestions without clear price expectations, which makes guests suspicious about the final bill.
• No connection with the kitchen or bar, meaning additional orders come in but are not executed smoothly.

In our frequently asked questions you can read how we address these three scenarios by default.

What it can deliver in practice

In the hospitality venues we support, we consistently see higher spending per table without harming the guest experience. The gains do not come from aggressive selling, but from consistently offering a relevant suggestion at the right moment. For a transparent overview of packages and features, see our pricing page. Additional background can be found at Hospitality Net on technology in hospitality.

Getting started with HorecaHub

Want to see what AI upselling in hospitality looks like for your venue? Explore the options on the pricing and packages page or schedule a no‑obligation conversation via contact. In a live demo we show what an average weekend of reservations could look like, at which moments a suggestion makes the difference, and how your team finally gets time back for the guests at the table.

Which seasons and occasions create extra leverage?

Certain periods of the year are better suited for upselling than others. Around Valentine’s Day, Easter, first communions, Father’s Day and Christmas the willingness to choose a package or upgrade measurably increases. AI uses this context so that reservations during those weeks automatically receive the most relevant package suggestion instead of the standard upsell. In this way you build a calendar of smart moments that repeats itself year after year without your team needing to think about it.

Local occasions such as city festivals, nearby concerts or major sporting events also offer upselling opportunities. A package with dinner and a pre‑event drink often works better on those evenings than a standard menu because guests are already in a going‑out mindset. The AI records which suggestions performed well and which did not, and automatically adjusts the selection next time.

How do you measure whether it works?

A good measurement of AI upselling looks not only at revenue per table, but also at the ratio between suggested and accepted offers. We report monthly on three KPIs: how many suggestions were sent, how many were accepted and how much additional revenue was actually generated. This way you see not only the end result, but also which suggestions you should refine and which you should remove.