Case study · Azzurro Club Vacanze · Email marketing & CRM
Hospitality · Lifecycle · CRM
Two email systems, one guest
The most valuable audience a hotel group has is the guests it already served, and the most expensive mistake is letting them come back through a platform that charges commission on a relationship the hotel had already earned. Email was the channel meant to stop that. Read as a system, it was two systems.
| Promotional (MailUp) | Transactional (booking system & app) | |
|---|---|---|
| Sender | one group address | one address per property |
| Register | informal, emoji, imperatives | formal, no emoji, no urgency |
| Personalisation | none | name, property, dates, party size, loyalty credit |
| Job | generate a lead for the call centre | run the stay and the loyalty loop |
The promotional side went out from one group address through MailUp: seasonal offers, one property or the whole group, written in the second person singular with emoji in the subject line, built on four levers that never changed — a time hook ("September is the new August"), an experiential promise, a stack of imperatives, and scarcity. The transactional side lived in the booking system and the app, with a different sender per property, a formal register, and something the promotional side never had: the guest's name, their property, their dates, the size of the party, their loyalty credit.
The lifecycle, as it actually ran
- Cold opt-in list → seasonal offer emails, no personalisation at all.
- Quote not converted → two dedicated flows: a "price freeze" that re-confirmed the old quote's conditions in full, and a re-activation that quoted back the exact period and party size the guest had asked for the year before. This was the sharpest piece of the whole programme.
- Booked → pre-check-in data collection and menu pre-ordering through the app.
- Checked out → loyalty credit plus a stay questionnaire.
- Loyal guest → a "customer week" with earned access ("you are receiving this because you qualified"), a 25% welcome-back discount on a second stay, and a 10€ referral for bringing a friend.
The best sequence in the group was the cross-sell cascade. The thank-you email at the end of the summer season closed the sea cycle and, in the same message, sold autumn weekends with theme parks and reminded the guest that the ski price freeze expired on 30 September. Three offers stacked on a contact who had just walked out of the hotel, at the one moment their goodwill was highest.
How the database was segmented
Five axes, of very different maturity: season and product (Riviera in summer, mountains in winter, with spring bridges and an autumn tail deliberately covered), property — each single-hotel email carried that property's own phone number so the call landed where the availability was, party composition, mostly families with children but explicitly broadened in the mountains to couples and groups of friends, lifecycle stage as above, and price tier, with the same database worked at 1,599–2,399€ a room in August, 999€ a week in September, and 139–299€ tickets for bridges and snow.
What was wrong with it, and what I would change
- The data was in the wrong system. Everything needed to personalise a promotional email — which property, which dates, how many people, how much credit — existed in the transactional side and was never used by the promotional side. That is the single biggest miss, and it is an integration problem before it is a marketing one.
- No email could close a sale. Even the fixed-price offer ended in "request a quote": every campaign fed the call centre. It fits a family audience that wants to speak to a person, but it makes revenue dependent on phone capacity and hides where the funnel actually leaks.
- Pressure instead of testing. The same spring creative went out four times in six weeks. There was no subject line test, no holdout, no frequency cap by segment: volume was doing the work that segmentation should have done.
- Two voices, sometimes in one email. The promotional register switched between informal and formal mid-message, with recurring typos in the templates. Small, but it is the kind of detail that tells a guest which emails are automated.
What I cannot show here: database size, open and click rates and the revenue attributed to email are in the group's systems and I did not keep a copy. The structure above is reconstructed from the campaigns themselves.