Transforming the digital omnichannel-marketing for the Telefonica Group
Role: Technical Product Owner Lead · 2023–2024
Customer data scattered across nine touchpoints. A unified CDP brought 21% cart recovery and 100K+ personalized messages.

- Cart recovery rate
- 21%
- Personalised messages
- 100K+
- Unified customer data
- 9 touchpoints
Team
- Team
- Tungi Dang
- Technical Product Owner Lead
Nine versions of the same customer
Telefónica Germany serves one of Europe's largest mobile customer bases. Behind that, customer data lived in nine systems — web, app, email, SMS, retail stores, call centres — each with its own record, its own logic, its own blind spots. Marketing couldn't personalise because no system knew the whole story. A support agent couldn't see what sales had promised. And every abandoned shopping cart simply stayed abandoned, because nothing downstream knew it existed.
I led the platform that had to turn nine partial customers into one real one.
The call: earn the platform with one legible win
Customer-data programs usually die of abstraction. "A single customer view" is a slide, not a result, and after a year of integration work with nothing visible, sponsors stop showing up. So the first delivery decision was scope: point the platform at the most concrete leak in the business — abandoned carts — and prove the whole chain on it. Not because carts were the biggest strategic prize, but because they were the most measurable one, in euros, within a quarter. That win would buy the patience the longer unification work needed.
Adobe Real-Time CDP unified the nine sources into one profile. Journey Optimizer handled orchestration, Experience Manager kept content consistent, Adobe Target ran the experiments.
The discipline was in what we didn't send
Germany is where marketing automation goes to get fined, so GDPR-compliant consent wasn't a checkbox — it was the operating constraint. Within it, restraint beat volume. Customers who dropped off mid-purchase got a coordinated sequence across web, email, SMS, and app, each message timed by the system rather than fired on a schedule. If they called in or walked into a store instead, the agent saw the cart in real time and picked up the conversation where it stopped.
In the first year: over 100,000 personalised messages, and 21% of cart abandoners returning to complete their purchase within 30 days. Sending more would have produced less — the recovery rate was a targeting result, not a reach result.
The whose-numbers-are-right phase disappeared
The quieter change was organisational. Marketing, BI, and Technology moved into one shared data environment: same profiles, same experiment results, same real-time analytics. Meetings that used to open with twenty minutes of reconciling dashboards started with the decision instead. Experimentation became routine rather than a special project, and offline signals from call centres and retail fed back into the same loop.
I established the consent management standards, data transparency protocols, and accountability lines across Marketing, IT, and Data Privacy that made all of this durable — personalisation only scales when customers can see it behaving well.
What changed
- 21% cart recovery within 30 days of abandonment; 100,000+ personalised messages in year one.
- One customer profile across all nine touchpoints, activated in real time from web to retail to call centre.
- A running experimentation culture on shared numbers, not a one-time integration project.
Governance and data-unification patterns from this work connect to The AI-native Platform Playbook.
21% cart recovery rate. 100K+ personalised messages. 9 touchpoints unified customer data.
