What we build
Every layer of the guest journey — personalised on one ledger.
Each capability ships as a production component — integrated to your PMS, CDP, and comms stack, measured against real revenue, and documented for your DPO.
Unified CDP across properties
PMS, CDP, loyalty, and comms history stitched into one guest record across every property — so the fourth stay continues the third, regardless of brand or market.
Preference learning, not guesswork
Room type, F&B, wellness, pace, and language encoded as embeddings that improve stay by stay — so every touchpoint is picked by fit, not by static segment.
Pre-arrival journey automation
F&B prefs confirmed, room readied against specific history, arrival windows set against flight timing — orchestrated from booking through welcome message, by language.
In-stay upsell by context
Weather, occupancy, folio position, and itinerary watched in real time — the right offer fires at the right hour, not a blanket mid-stay survey that reads the same to everyone.
Folio-aware messaging
Every message is aware of spend, outstanding balance, and transactional context — no upgrade pushes to guests already upgraded, no spa comms post-spa, no noise.
12+ APAC languages, native
EN, ZH, JA, KO, TH, ID, MS, VI, TL, HI, plus locale splits (zh-HK, zh-TW). Tone and register matched per brand and per channel — not a translation layer bolted on.
Guest journeys we personalise
One engine, every stage of the stay.
Same preference graph, context engine, and language layer — tuned per segment. Shared features, per-brand tone, portfolio-level measurement. Every journey below runs on the same stack; only the tone, cadence, and integration map change.
Pre-arrival readiness
Room prefs, arrival timing, F&B, and language all confirmed before the guest lands. Housekeeping and F&B both see the same brief.
In-stay upsell orchestration
Context-aware offers — weather-triggered spa, occupancy-aware upgrades, folio-aware F&B — routed by propensity, not by calendar.
Post-stay advocacy loops
Loyalty-tier-aware thank-yous, review invitations, and return-stay nudges in the guest's preferred channel — warm without being templated.
Multi-property journeys
Cross-property recognition so a repeat guest at one brand arrives recognised at another — preferences carried forward, not asked again.
MICE + group segmentation
Per-delegate personalisation at group scale — arrival staggering, dietary capture, language-matched welcome flows, NPS visibility by sub-cohort.
Loyalty-tier automation
Black, Platinum, and Gold journeys each run their own message cadence, upgrade policy, and channel mix — shared stack, per-tier experience.
Model families we deploy
No single model covers every guest. So we ensemble.
Each model family covers a distinct decision — combining their outputs gives you coverage, resilience, and an audit trail a rules engine alone can't match.
Guest behaviour encoded into a 128-dimension preference vector — F&B, wellness, room type, pace, language. Learns from every stay and updates against portfolio baseline.
Per-guest, per-offer likelihood of acceptance, calibrated on real take-rate data. Drives upgrade, spa, F&B, and package offers — with confidence surfaced per decision.
Learns the best send hour and channel per guest segment. Balances explore vs exploit so new cohorts are learned, not force-fit to last year's template.
Classifies each guest's preferred language from prior messages, loyalty profile, and stay history — with locale granularity (zh-HK vs zh-TW) and tone preference.
Data sources wired into every journey
Every signal that makes the message land — integrated.
Pulled in parallel, normalised into a single guest schema, consented-tagged before any model consumes them.
Explainability, not just output
A personalised message isn't trusted unless the operator knows why it fired.
Every send is accompanied by top-feature reasons, model and consent provenance, and a customer-facing rationale where required — generated at decision time, indexed for audit, and available in the languages your guest speaks.
- Top-feature contributions logged per decision
- Full consent + model-version provenance
- Guest-facing rationale where rules require
- Aligned to GDPR, PDPA, PIPL, CCPA, APPI
Why Axccelerate for guest personalisation
Not a campaign tool.
A personalisation stack.
A campaign tool gives you templates. Our stack gives you a unified guest graph, preference learning, context triggers, consent-first routing, and portfolio measurement — the infrastructure a real hotelier actually needs.
Pricing
Priced to the portfolio, not the guest volume.
Personalisation deployments are custom — we scope against your properties, CDP, and integration map before quoting.
Glossary
The vocabulary behind every touchpoint.
A quick reference for the acronyms that show up in guest personalisation — the terms your commercial team, DPO, and operators will all use.
- CDP
- Customer Data Platform
A unified guest database that stitches PMS, booking, loyalty, and campaign history into a single profile with identity resolution and consent tracking.
- PMS
- Property Management System
The operational core of a hotel — reservations, room assignment, rates, housekeeping, and billing. Opera, Mews, and Cloudbeds are common examples.
- Folio
- Guest folio · stay account
The running account of every charge a guest incurs during a stay — room, F&B, spa, retail. Folio-aware messaging avoids pushing what the guest has already bought.
- Upsell
- Revenue-lift sale
Any paid add-on beyond the booked rate — room upgrade, late checkout, dining, wellness. The main ancillary-revenue lever in hospitality personalisation.
- Propensity score
- Likelihood-of-acceptance score
A per-guest, per-offer probability of a positive outcome. Models are calibrated against actual take-rate, not just historical opens or clicks.
- Segment vs cohort
- Static group vs behavioural bucket
Segments are rule-based (Gold tier, 2+ stays). Cohorts group guests by a time-bound shared behaviour (booked in week X) and power bandit-style learning.
- Loyalty tier
- Recognition level
A hierarchical status bracket (Black, Platinum, Gold, etc.) that drives recognition rules, upgrade priority, and journey cadence in the personalisation engine.
- Consent-based marketing
- Opt-in outreach
Messaging that only reaches guests who have given explicit, versioned consent for the specific channel and purpose — the GDPR / PDPA baseline.
- Dwell time
- Time-on-property signal
How long a guest has been in-stay when a trigger fires. Used to avoid cold-open upsells and to time high-intent comms (spa mid-stay, dining at sunset).
- RFM
- Recency · Frequency · Monetary
A classical three-axis segmentation still useful as a baseline. Personalisation models extend this with preference vectors, language, and channel data.
- LTV
- Guest Lifetime Value
The projected total spend of a guest across a portfolio over a set horizon. Drives how aggressively the engine invests in recognition and upsell offers.
- Channel preference
- Preferred delivery path
The channel (email, WhatsApp, in-app, SMS) a guest is most responsive to, learned from open / action rates and explicitly honoured by the engine.
- Preference centre
- Guest-facing consent panel
A guest-controlled interface where channel, cadence, and topic opt-ins are managed. Every change is versioned and propagated across the platform.
- Drip sequence
- Cadenced message series
A scheduled, conditionally branched series of messages — pre-arrival, in-stay, post-stay — each gated by guest state, preference, and propensity.
Your guest graph, engineered.
30-minute scoping with a senior engineer and a hospitality-systems operator. You'll leave with an integration sketch, model plan, and realistic timeline — not a sales pitch.