Jan Hendrik Fleury - j.fl[email protected] A practical approach to use big data and customer technology for customer centricity

No content

1. The customer technology ecosystem 2. Types of Customer Data Platforms 3. Case FD Media Group 4. How to start 5. Takeaways Agenda

BEHAVIOR INTEREST PERSONALITY INTENT DEM OGRAPHICS BEHAVIOR INTEREST PERSONALITY INTENT DEM OGRAPHICS DATA REACH collect sources 0 + 1ST PARTY DATA 2ND PARTY DATA 3RD PARTY DATA DATA PRIVACY trust and transparency DATA ONBOARDING high data accuracy 360° CONSUMER VIEW single source of truth ENRICHMENT predictive & generative ORCHESTRATED ACTIVATION micro targeting ENTERPRISE DATA LAYERS ANONYMOUS USERS IDENTIFIED USERS AUDIENCE SEGMENTATION DATA SCIENCE CONTENT PERSONALIZATION DIRECT MARKETING TARGET ADVERTISING content audiences message time location frequency remarketing GDPR DATA DECISIONS ENGAGEMENT Customer technology supplies and needs data…

BEHAVIOR INTEREST PERSONALITY INTENT DEM OGRAPHICS BEHAVIOR INTEREST PERSONALITY INTENT DEM OGRAPHICS DATA REACH collect sources 0 + 1ST PARTY DATA collected and owned by company 2ND PARTY DATA partner data shared with company 3RD PARTY DATA owned by others DATA GOVERNANCE DATA PRIVACY trust and transparency DATA ONBOARDING high data accuracy 360° CONSUMER VIEW single source of truth ENRICHMENT predictive & generative ORCHESTRATED ACTIVATION micro targeting FULL FUNNEL DATA STREAMING EVENTS & PURCHASES SOCIAL MEDIA APP & WEB EMAIL, PUSH & ECOMMERCE MONITORING DATA AND COST WORKFLOW ORCHESTRATION BUSINESS INTELLIGENCE PERFORMANCE MEASUREMENT CONSENT LEGITIMATE INTEREST privacy management transparency and consent ENTERPRISE DATA LAYER identity resolution tag management content classification metadata ANONYMOUS USERS IDENTIFIED USERS account email address purchase AUDIENCE SEGMENTATION SMART DATA SCIENCE pCLT modelling lookalike modelling behavior analysis propensity modelling conversion modeling engagement modelling churn prediction automated testing CONTENT PERSONALIZATION personalization and recommendations MARKETING AUTOMATION campaigns flows TARGET ADVERTISING programmatic addressable contextual content audiences message time location frequency remarketing GDPR PDCA … and requires a robust ecosystem

Streaming Batch Collect Process Store Data Sources Data Destinations Customer Data Platform Serverless BigQuery Storage Cloud Storage Cloud Dataflow Pipelines Ingest Cloud Pub/Sub Storage Cloud Storage Messaging (Push Email / SMS) Paid marketing channels Websites & Apps Privacy controls Identity Resolution Campaign Personalization Insights Modelling Clean data Raw data Curated data Web & App Stock Content metadata CDP logs ERP & Shipment Marketing & Ad Server logs Customer Service Orchestration - Cataloging - Governance - Security - Monitoring & Alerting Type 1 - Packaged suite CDPs Suite vendors have biggest chunk of functions

Collect Process Store Data Sources Data Destinations Modular CDP Components Stock Web & App Content metadata BigQuery Storage Cloud Storage Cloud Dataflow Pipelines Ingest Cloud Pub/Sub Streaming Batch Storage Cloud Storage Messaging (Push / email / SMS)) Paid marketing channels Website & Apps Identity Resolution & Privacy Campaign & Delivery Personalization & Modelling Insights Clean data Raw data Curated data CDP logs ERP & Shipment Marketing & Ad Server logs Customer Service Orchestration - Cataloging - Governance - Security - Monitoring & Alerting Serverless Type 2 - Headless composable CDPs - Best of Breed Hybrid GCP and packed SAAS/ISV

Collect Process Store / Analyse Data Sources Data Destinations BigQuery Storage Cloud Firestore Cloud Dataflow Pipelines Ingest Cloud Pub/Sub Storage Cloud Storage Messaging (Email / SMS / Push) Websites & Apps Marketing Channels Orchestration - Cataloging - Governance - Security - Monitoring & Alerting Activate Recommen- dations AI Cloud Endpoints App Engine Presentation Cloud Functions Serverless Customer Service Stock Web & App Content metadata Batch CDP logs ERP & Shipment Marketing & Ad Server logs Insights Vertex AI Type 3 - Full Headless CDP Majority of Crystalloids’ clients on native GCP products, limited use of 3rd party SAAS / ISV

Headless / composable in more detail Increasingly popular approach without packaged CDP ● The cloud data warehouse GCP BigQuery acts as the data foundation ● Actions and insights are created on all levels, not limited to customer data ● Activation of data directly from the cloud data warehouse environment ● A combination of cloud-native services and specialized tools and frameworks

Functionality comparison

When to choose headless? If you have or plan to establish a data warehouse on BigQuery ● Advanced and more complex use cases ● Not want to duplicate data and recreate business rules in different environments ● Optimally leverage existing investments in a cloud data warehouse ● Unlock the potential of non-customer data ● Reduce vendor lock-in of systems and applications ● If the cost of the packaged CDP grows 10X while your business grows 2X ● Dedicated tools to connect data and tools easily, bringing flexibility of choice ● Flexibility in on- and offboarding point solutions in the future

het Financieele Dagblad, BNR, Company.info Democratizing data, activation & measurement Goals ○ Future proof setup of first-party data collection and profiling ○ Acquisition and retention tactics based on data-driven targeting ○ Advertisements based on context, customer interest and 1p data ○ Article tagging for content optimization ○ Democratize data for self-service analytics enterprise level ○ Governed Generative AI platform (WIP) Solutions ○ Unified modular martech stack on GCP centralizing enterprise data, analytics, activation and cross channel measurement in a CDP. ○ Data strategy design and implementation to support functional and technical data driven transformation. ○ Growing data literacy with inhouse training. Result ○ Mitigation of EUR 2.000k revenue decrease at risk ○ 200% increase CPM ad sales ○ 30% efficiency increase marketing and data analytics teams ○ Comprehensive insights in ROI of campaigns ○ Insights is use cases for Gen AI Customer cases on Crystalloids and on Google Cloud website

Architecture Collect Process Store Analyse Consume Source data Consumers GTM Web container Business Intelligence Marketing, sales, customer services GTM Server side Website Data Layer AppEngine Management Cloud Pub/Sub Ingest Pipelines Cloud Dataflow Google Analytics Cloud Storage Tableau Analysts Activate Google Ad Manager Google Ads User / Visitor GitLab Governance Cloud Composer Cloud Build Data layer Selligent Marketing Cloud Looker Meta (Facebook) Selligent Optin, profile, performance Customer Signals FD Profile data SAP Profile & order data SalesForce Datahub tables Engagement-, Churn data Company Info Machine Learning Vertex AI Article Content text, images, tags Storage BigQuery Explore / Analyse BigQuery Cloud Storage Storage Select / Trigger Cloud Functions data - preprocessing model - training - registration - serving - scoring - evaluation - monitoring Keycloak Login events Website Article & Subscription events Trusted data AI Publishers & journalists Cloud Composer Web & App

Start with defining use cases

Workshop Gathering input from business teams together Prioritize Focus on user stories that add the most value Refine Refined top priority user stories and added them the next sprint and backlog Make sense of opinions Prioritize and refine user stories

Architecture Target architecture start and end state SCRUM team Set up team Product Owner, SCRUM Master, Architect Developer, Data Scientist Build Stand ups, develop, demo, retrospective Design, team up, build and show Developing the solution incrementally

6. Build selected user stories 8. Test, evaluate, improve 9. Show and tell to stakeholders 10. Built new use cases Path to realizing benefits 4 - 6 Days in total Development sprints Discovery sprint 1. Co creation backlog of user stories 1. Refinement and prioritization of the backlog 1. Select stories to develop with management incl. high level business case 1. Technical discovery + target architecture 1. Detailed estimation of work (incl. estimation workload for client ) + detailed business case < Go-no go decision > 6. Build selected user stories 8. Test, evaluate, improve 9. Show and tell to stakeholders 10. Built new use cases Development Sprints Results after 4 - 6 weeks

Takeaways ● A CDP is a combination of functions and not a tool ● Balance between Packaged, ISVs and Headless ● Don’t boil the ocean; leave the legacy systems as they are ● Build as much as you can on the cloud with cloud native solutions ● Add the 3rd party components you might miss ● Boost ROI on ESP’s, ecommerce systems etc by feeding them with the right data and decisions ● Develop tech, skills and use cases incrementally ● Grow employee satisfaction this way

Download your free copy