Skip to content

How AI agents are changing hotel upselling: beyond the front desk pitch

Tips for hotels · 6 min read

In short

Hotel upselling works when the offer reaches the guest on the channel they already use, at the moment they are still deciding. A generic pre-arrival email and a front desk pitch both miss that window. Early check-in, late checkout, parking and room upgrades convert best when offered shortly before arrival, personalised to the booking rather than blasted to everyone.

The problem with traditional hotel upselling

Most hotel upselling happens in two places: a generic pre-arrival email and a front desk pitch at check-in. Both have problems.

The email blast goes to every guest with the same offer. Open rates are low. Personalization is nonexistent. The front desk pitch interrupts what should be a smooth arrival experience, and staff often skip it during busy periods.

The result: hotels leave significant revenue on the table. Industry data suggests the average hotel captures only 10-15% of potential ancillary revenue through manual upselling.

How AI changes the equation

AI agents change upselling from a one-shot event to a continuous, context-aware process across the entire guest journey.

Before arrival: The agent analyzes the booking (party size, room type, dates, special occasions) and sends personalized offers via the guest’s preferred channel. A couple arriving on an anniversary gets a spa package offer. A business traveler gets early check-in and airport transfer. A family gets connecting room upgrades and kids’ activity packages.

Timing matters: Offers go out when guests are most receptive. Research shows the 48-72 hour window before arrival has the highest conversion rate for room upgrades. Day-of-arrival works best for dining packages. During-stay prompts convert best for experience add-ons.

During stay: The agent monitors guest interactions for upsell signals. A guest asking about restaurant recommendations gets a dining package offer. Someone requesting late checkout gets offered a full extra night at a discount. Every interaction is an opportunity, handled naturally, not pushily.

Making it feel helpful, not pushy

The difference between good and bad upselling is whether the guest feels helped or sold to. AI agents excel here because they can:

  • Personalize deeply: using booking data, past stay history, and real-time interactions
  • Time precisely: sending offers when they’re most relevant, not on a batch schedule
  • Respect boundaries: recognizing when a guest isn’t interested and backing off gracefully
  • Provide value first: leading with useful information (arrival tips, local guides) before suggesting paid upgrades

When a guest receives a WhatsApp message with genuinely useful arrival information followed by a relevant upgrade offer, it feels like a concierge service, not a sales funnel. It also feeds the score: Booking.com states that its guest review score is weighted by recency since January 2025 (opens in a new window), so the experience you give guests this month moves your rating faster than it used to. An AI receptionist that answers questions and offers the upgrade in the same thread is doing both jobs at once.

Implementation: start with the highest-converting offers

Don’t try to upsell everything at once. Rank by typical conversion rate and start with the top three:

  1. Room upgrades, offered around 48 hours before arrival over WhatsApp, while the guest is still planning the trip
  2. Airport transfers (20-30% conversion for leisure travelers)
  3. Early check-in / late checkout (25-35% conversion when offered contextually)
  4. Dining packages (10-20% conversion during stay)
  5. Experience add-ons (5-15% conversion, highly seasonal)

Connect your AI agent to your PMS so it has real-time availability and pricing. Nothing kills conversion like offering an upgrade that isn’t actually available.

How Timo's AI receptionist sells at the right moment

Timo's AI receptionist offers extras inside a conversation the guest is already having, and only when the offer fits their question. The guest asks when they need to leave the room, Timo answers and offers a late check-out with it. Whatever the guest accepts is noted on the reservation, so the front desk sees it in the morning overview:

Daniel Brooks

Hotel Linden · room 108 · leaving on Sunday

WhatsApp

Good evening, what time do we need to leave the room on Sunday?

Good evening, Mr Brooks. Check-out is at 11:00. If you would like to stay longer, a late check-out until 14:00 is 25 €. Shall I add it for you?

Yes please, that suits us.

Done, your late check-out until 14:00 is noted on your reservation. The front desk knows.

20:15

Late check-out sold · 25 €Noted on the reservation
The offer came as the answer to the guest's question, not as a mass message

You set the offers and the rules for whom Timo offers them to and when. How it works for other services is shown under selling services.

What hotels are seeing

Properties running AI-powered upselling typically report:

  • 25-40% increase in ancillary revenue per guest
  • 3-5x higher conversion on personalized vs. generic offers
  • Zero additional front desk workload
  • Higher guest satisfaction (guests appreciate relevant suggestions)

The revenue impact alone usually covers the cost of AI implementation within the first quarter.

Frequently asked questions

What is hotel upselling?

Hotel upselling is offering a guest something beyond the room they booked: an upgrade, early check-in, late checkout, parking, a transfer, breakfast or a package. It is incremental revenue on a booking you already have, which is why it carries no acquisition cost and no OTA commission when it is sold directly by the property.

When is the best time to send an upsell offer?

Shortly before arrival, when the trip is real enough to plan and there is still time to say yes, and again at the moment a related need appears during the stay. The two windows that consistently underperform are the booking confirmation, which is too early, and the check-in desk, where the guest is tired and other guests are queueing behind them.

Which hotel upsells convert best?

The ones that solve a problem the guest already has. Early check-in for a morning arrival, late checkout for an evening flight, parking for a guest who is driving, an upgrade for a couple travelling on an anniversary. Relevance beats discount: an offer that matches the booking outperforms a bigger discount sent to everyone.

Does automated upselling annoy guests?

It annoys them when it is untargeted, repeated or interrupts a complaint. It does not when the offer is relevant, easy to decline and arrives once. The practical test is whether the message would still be useful if the guest said no, which is why the strongest upsell messages carry genuinely useful arrival information as well as the offer.

Do I need a PMS integration to automate upsells?

For anything personalised, yes. Without the reservation the assistant does not know the arrival time, the room type or whether the guest already paid for parking, so every offer becomes a generic blast. With PMS access it can target the guest whose booking actually implies the need, and write the accepted upsell back to the reservation.

About the author

Marek Didák (opens in a new window)

Marek looks after hotels that are getting started with Timo. He helps them decide when Timo replies and when their own people do.

All articles

We’ll show you Timo handling your guests’ questions

Book a meeting