01
The brief, in full
Network, marketing and customer care each had their own numbers for the same customers. Care could not see a local outage, marketing could offer an upgrade where there was no coverage, and every report began with an argument about whose figure was right.
The brief: give network, marketing and care one version of the truth.
02
Our approach
We started with definitions, not dashboards: one metric catalogue agreed by all three teams, each metric with an owner and its lineage. The shared data layer was built to serve those definitions, and every dashboard reads from it.
On that layer sit a care assistant that can see network status, with a hand-off to an agent, and registered, consented messaging that knows where coverage is.
How it ran
M2M4M6M8M10
-
Weeks 1–6Definitions
The metric catalogue, agreed by network, marketing and care.
-
Months 2–6Platform
The shared data layer and the dashboards that read from it.
-
Months 5–9Assistant
The care assistant, its eval set and the hand-off to an agent.
-
Months 7–12Messaging
Registered messaging and the preference centre, checked against coverage.
03
What we built
A shared data layer and metric definitions used by network, marketing and care, feeding a care assistant and registered, consented customer messaging.
What it left behind
- 01Data platform and metric catalogue
- 02Care assistant and eval set
- 03Preference centre
- 04Operational dashboards
04
What changed
The layer is built so network, marketing and care read the same figures from the same place: the care assistant answers with live network status and hands over to an agent when it should, and offers are checked against coverage before they are sent.
This programme is confidential: its figures are shared only under NDA.
How it is measured ›