CX-10 · Customer Data & Personalization

📊

Process Flow Diagram (BPMN)

📋

L4 Process Steps

Step	Step Name	Role / Swim Lane	System	Input	Output	KPI	Dec?	Exc?
Phase 1 1.1	Ingest booking & transaction data from PSS	Data Engineer	Amadeus Altéa PSS	Nightly batch feed of PNR, ticketing, and ancillary purchase records	Raw booking event stream loaded to data lake	Data pipeline SLA: 100% of prior-day records ingested by 04:00 local	N	N
1.2	Collect loyalty programme interaction events	Data Engineer	Salesforce Marketing Cloud	Email opens, loyalty portal logins, miles accrual and redemption events	Loyalty event stream appended to data lake	Event capture rate ≥99% of Salesforce Marketing Cloud triggers per day	N	N
1.3	Capture web & mobile behavioural data	Digital Analytics Analyst	Adobe Analytics	Clickstream, search, seat-map interaction, and app session events	Behavioural event stream with session IDs written to AWS S3	Tag coverage ≥98% of key conversion pages; session data latency ≤2 hours	N	N
1.4	Validate data completeness across all sources	Data Quality Engineer	AWS S3 / Redshift	Ingested event counts vs expected record volumes per source	Completeness scorecard; pipeline held or released	Overall data completeness ≥95% before downstream processing proceeds	Y	Y
Phase 2 2.1	Cleanse and deduplicate customer records	Data Steward	Adobe Experience Platform	Raw multi-source event streams from data lake	Standardised, deduplicated customer record set	Duplicate record rate ≤0.5% post-cleanse; address standardisation accuracy ≥97%	N	N
2.2	Verify PII consent and GDPR / CCPA compliance	Privacy & Compliance Analyst	OneTrust	Customer consent records, marketing opt-in flags, data retention schedules	Consent-cleared record set; non-consented records quarantined	100% of records processed against current consent status before any marketing activation; consent sync latency ≤24 hours from collection event	Y	Y
2.3	Enforce data retention and right-to-erasure rules	Privacy & Compliance Analyst	OneTrust	Erasure requests from customer portal, regulatory notices	Deletion confirmation across all downstream systems within SLA	Right-to-erasure requests fulfilled within 30 days per GDPR Article 17; 100% completion rate	N	Y
Phase 3 3.1	Resolve identity and stitch cross-channel profiles	Data Scientist	Adobe Experience Platform	Consent-cleared, cleansed records with email, loyalty ID, device ID, PNR	Unified customer profiles with deterministic and probabilistic identity links	Identity match rate ≥80% of known travellers; profile stitching latency ≤6 hours from data ingestion	Y	N
3.2	Compute customer lifetime value and tier segments	CRM Analytics Manager	AWS Redshift	Unified profiles enriched with 24-month transaction history	CLV score, predicted annual revenue, and tier classification per customer	CLV model R² ≥0.72; tier classification refreshed weekly; top-decile CLV customers identified for proactive retention	N	N
Phase 4 4.1	Run customer segmentation models	Data Scientist	AWS Redshift	Unified profiles with CLV, travel frequency, route preferences, ancillary purchase history	Customer segment labels (e.g., Leisure Infrequent, Corporate Road Warrior, Premium Aspirant)	Silhouette score ≥0.55 for k-means clusters; segment refresh cycle ≤7 days	Y	N
4.2	Build propensity scores for ancillary and upgrade offers	Data Scientist	Adobe Experience Platform	Segment labels, historical ancillary purchase events, seat-map interaction data	Per-customer propensity scores: upgrade likelihood, seat upsell, lounge day-pass, travel insurance	AUC ≥0.75 per propensity model; score refresh ≤48 hours before scheduled departure	Y	N
4.3	Score churn risk for loyalty programme members	CRM Analytics Manager	Salesforce Marketing Cloud	Loyalty transaction recency, frequency, redemption gap, competitive route exposure	Churn risk tier (High / Medium / Low) and recommended retention intervention	Churn model recall ≥70% on held-out validation set; High-risk members flagged ≥60 days before predicted lapse	N	N
Phase 5 5.1	Generate personalised pre-trip offer packages	CRM Campaign Manager	Salesforce Marketing Cloud	Propensity scores, segment labels, PNR departure date, fare class booked	Personalised offer payload (upgrade, seat, lounge, hotel partner) per customer	Offer relevance score ≥70% (measured by click-through proxy); ancillary attach rate uplift ≥8% vs control	Y	Y
5.2	Activate offers across email, app push, and web	Digital Marketing Specialist	Braze	Personalised offer payloads, customer channel preference flags, optimal send-time scores	Delivered personalised messages across email, push notification, and website homepage	Email open rate ≥28%; push notification opt-in rate ≥45%; web personalisation CTR ≥4%	N	N
5.3	Serve real-time personalisation at airport touchpoints	Digital Product Manager	SITA Airport Management System (AMS)	Customer profile, check-in event trigger, real-time upgrade inventory from Amadeus Altéa PSS	Targeted upgrade or ancillary prompt displayed at kiosk or agent terminal	Kiosk upgrade conversion rate ≥5% for High-propensity customers; offer display latency ≤500ms from check-in event	Y	Y
Phase 6 6.1	Track offer redemption and revenue attribution	CRM Analytics Manager	AWS Redshift	Offer delivery logs from Braze and SITA AMS, ancillary revenue from Amadeus Altéa PSS	Offer conversion rate, incremental revenue per segment, cost-per-acquisition	Incremental ancillary revenue per personalised journey ≥€12 vs control; attribution model updated weekly	N	N
6.2	Analyse A/B test results and model performance	Data Scientist	Tableau	Conversion data, control vs treatment group outcomes, propensity model predictions vs actuals	Performance report; decision to retrain models or adjust segmentation thresholds	A/B test statistical significance ≥95% before rolling out winning variant; model performance reviewed bi-weekly	Y	N

📋

Process Attributes

Identification

Process IDCX-10

L1 DomainCustomer Experience & Loyalty

L2 ProcessCustomer Experience Operations

L3 NameCustomer Data & Personalization

L4 Steps17 across 6 phases

Decision Gates8 (all with iteration loops)

Exceptions5 documented

Swim Lanes (Roles)

Data Engineer

Digital Analytics Analyst

Data Quality Engineer

Data Steward

Privacy & Compliance Analyst

Data Scientist

CRM Analytics Manager

CRM Campaign Manager

Digital Marketing Specialist

Digital Product Manager

Systems & Tools

Amadeus Altéa PSSSalesforce Marketing CloudAdobe AnalyticsAWS S3 / RedshiftAdobe Experience PlatformOneTrustAWS RedshiftBrazeSITA Airport Management System (AMS)Tableau

Key Performance Indicators

Ingest booking & transaction data from PSSData pipeline SLA: 100% of prior-day records ingested by 04:00 local

Collect loyalty programme interaction eventsEvent capture rate ≥99% of Salesforce Marketing Cloud triggers per day

Capture web & mobile behavioural dataTag coverage ≥98% of key conversion pages; session data latency ≤2 hours

Validate data completeness across all sourcesOverall data completeness ≥95% before downstream processing proceeds

Cleanse and deduplicate customer recordsDuplicate record rate ≤0.5% post-cleanse; address standardisation accuracy ≥97%

Verify PII consent and GDPR / CCPA compliance100% of records processed against current consent status before any marketing activation; consent sync latency ≤24 hours from collection event

Enforce data retention and right-to-erasure rulesRight-to-erasure requests fulfilled within 30 days per GDPR Article 17; 100% completion rate

Resolve identity and stitch cross-channel profilesIdentity match rate ≥80% of known travellers; profile stitching latency ≤6 hours from data ingestion

Airline-Specific Risks & Pain Points

Altéa PNR schema changes on major releases can silently break field mappings, causing downstream profile gaps until the next data audit cycle

Loyalty tier change events arrive asynchronously from PSS; without an event-ordering layer, a customer can receive a downgrade communication before the upgrade is confirmed

iOS ATT (App Tracking Transparency) and Safari ITP limit cross-device identity stitching, reducing identifiable behavioural profiles by ~20–30% on Apple devices

GDS-originated bookings (Sabre, Travelport, Amadeus GDS) often arrive with partial passenger profile fields; missing DOB or contact email breaks identity resolution

Passengers booking via third-party OTAs (Expedia, booking.com) often have name/email variations that resist standard matching rules, inflating duplicate counts

Consent revocation must propagate to Salesforce Marketing Cloud, Adobe Experience Platform, and Braze within the 72-hour GDPR window — cross-system latency routinely misses this SLA without automated orchestration

Inputs / Outputs

Primary InputNightly batch feed of PNR, ticketing, and ancillary purchase records

Primary OutputPerformance report; decision to retrain models or adjust segmentation thresholds

←PreviousCX-09 · Special Services Request (SSR) Management NextCX-11 · IRROP Passenger Recovery→