Case study
Conversational selling for hospitality: from QR to phone call
Case study: a hotel-room QR becomes a virtual host who profiles each guest in their language and ends in a real AI phone call in under 30 seconds.
A guest scans a QR code by the elevator. Minutes later their phone rings, and the voice on the line speaks Ukrainian, because that's what they picked on the first screen.
We built this for an international hotel chain with resorts in the Caribbean: a conversational selling experience that starts at a QR code in the room or lobby and ends with a real phone call from an AI voice agent, under 30 seconds after the guest submits a short form. In between: a virtual host on video, a quiz that builds each traveler's profile, a different micro-pitch per profile, and five languages where the one you pick at the start follows you all the way to the call. Our own pipeline produced and deployed the whole thing in hours, not weeks.
The challenge: a captive guest and a PDF nobody opens
Hotels have something almost no other business gets: the customer physically inside, with time to spare. What that customer usually finds is an acrylic stand with a QR that opens a PDF.
A PDF asks no questions. It can't tell a family with two kids from an anniversary couple. It's written in English, which does little for the guest who flew in from Tbilisi. And the front desk isn't going to pitch in Georgian at 11 pm.
The brief wasn't "add AI to the hotel." It was: turn that moment of attention into sales.
The experience, in three acts
Act one: the host. The QR opens a language picker: Spanish, English, Chinese, Ukrainian, Georgian. Pick one and she appears, a virtual host with a local accent and subtitles, welcoming you on video the way someone would at the door. Not a menu of buttons. A person talking to you.
Act two: the conversation. She asks what kind of trip this is. Here's the detail guests never consciously register but always feel: the answer options surface over the video at the exact instant she speaks the question. The sync is word-level, not "roughly when the clip ends." Your answers build a traveler profile, and she follows with a short video pitch made for that profile. We produced five variants; a family hears a different pitch than a couple.
Act three: the call. A closing video, a short form, then the payoff: the guest's phone rings in under 30 seconds. On the line, an AI voice agent that already knows the language, the profile and what caught their interest. No "thanks, we'll be in touch." A call, right now.
Pick Georgian on the first screen and the call arrives in Georgian. That thread never breaks.
How it was built, without opening the hood
Seven video pieces per language: one hook, five micro-pitches, one closer. Across five languages: 35 videos. Traditional production would mean weeks of casting, studio time, per-language voice talent and post. Our pipeline produced and deployed everything in hours.
The machinery is ours and we don't publish it. We will tell you what it allows: changing a script, adjusting a pitch or adding a language without mounting another video production. The experience iterates like software, not like a TV commercial.
What it left running
We don't publish client business metrics. Be wary of anyone who does without permission. What is ours, and measurable:
- 35 videos produced: 7 pieces across 5 languages
- 5 languages end to end: Spanish, English, Chinese, Ukrainian, Georgian
- Phone call under 30 seconds after the form
- Produced and deployed in hours, not weeks
Pipeline capacity figures, not client results.
The qualitative part matters too: the guest never repeats information, never gets switched to English halfway through, and it all ends where sales actually close: in a voice conversation.
FAQ
What is a conversational selling experience? A funnel where the prospect talks instead of reads: video that speaks to them, questions answered with a tap, a voice call at the end. It replaces the PDF with something closer to your best salesperson at work.
Is the host a filmed actress? She's an AI avatar we produce, with a local accent and subtitles. That's how five pitch variants could exist in five languages; with actors, that volume of footage would be unworkable on time and cost.
Who makes the phone call? An AI voice agent. It calls under 30 seconds after the form, in the language the guest chose at the start, carrying the quiz profile as context. It doesn't start from zero; it continues the conversation.
Which languages does it support? This case ran in Spanish, English, Chinese, Ukrainian and Georgian, with the chosen language traveling from first screen to phone call. Adding a language takes hours, not another production.
How long does deployment take? Hours. The pipeline produces the videos, syncs them with the quiz and puts the experience live the same day. Iterating a script or a pitch moves at the same speed.
Does this only work for hotels? No. Hospitality set an ideal stage, a customer already present with time to spare, but the pattern applies wherever there's a moment of attention and something to sell. At mayalab we start with the as-is: you walk us through how you sell today, and we design the experience on top of that.
Book a demo
Want to walk the full experience, QR to phone call? Write to [email protected] and book a demo. We'll run it in your language.