Case studies Hospitality

3x upsell revenue from an automated guest journey

The system communicates with each guest from booking to a week after checkout: practical information, timed offers, and a mid-stay satisfaction check. When a guest is unhappy, management is alerted while the guest is still on site.

One guest, one stay

  1. Booking
  2. Pre-arrival
  3. Check-in
  4. In-stay guide
  5. Day two check
  6. Escalation
  7. Checkout
  8. Follow-up

Alerted to hotel management

Room 412 rated 2 of 5 on day two. The guest is still on site, so the problem can still be fixed.

ENBooking confirmed
  • Every guest contacted from booking to a week after checkout
  • Offers proposed at the point the guest is most receptive
  • Low ratings escalated while the guest is still on site
  • Each guest answered in the language they wrote in, around the clock
Industry
Hospitality
Service
AI agents and intelligent automation

ContextWhat the client was facing

Across three properties, guest communication was manual, inconsistent, and reactive:

  • Reception fielded the same questions repeatedly: parking, WiFi, check-in times, directions
  • Additional services were offered only if a guest happened to ask at the desk
  • Dissatisfied guests said nothing on site and wrote the review after they got home
  • Review requests were sent inconsistently, so satisfied guests often left without leaving one
  • Guests arrived from several markets, and out-of-hours coverage did not span their languages

SolutionWhat we built

We built an automated guest journey covering the full stay, integrated with the group's existing booking system:

  1. Timed, personalised messaging Pre-arrival details, check-in information, and a real-time digital in-stay guide.
  2. Offers matched to guest profile Room upgrades, spa, dining, and transfers proposed at the point the guest is most receptive.
  3. Mid-stay satisfaction check A short rating request on day two or three.
  4. Escalation on low scores A low rating alerts hotel management immediately, so the problem is resolved while the guest is still on site.
  5. Post-stay follow-up Thank you, direct review link, and a return offer, sent on a fixed schedule.
  6. Automatic language detection Each guest is answered in the language they wrote in, around the clock.

OutcomeResults and value delivered

3×

upsell revenue per stay, from offers timed to the guest

62%

of low-rated stays recovered on site, before the guest left the property

47%

more reviews collected, with average score rising from 4.1 to 4.4

58%

fewer routine enquiries reaching reception

  • Consistent guest experience across all three properties, regardless of shift or season
  • Every escalation logged, giving management visibility into recurring complaints by property

Tripled upsell revenue and 47% more reviews, driven by a system that knows when to sell, and when to step aside and fetch a human.

Technology stack

  • Python
  • PostgreSQL
  • PMS integration
  • Anthropic Claude
  • AWS

Next

68% of property enquiries resolved with an AI voice agent

Real estate

Start a project