How Vodafone business (VOIS) build a platform to enable and execute analytics models to create multiple tactics for increasing the efficiency for, on call processing at call centre

Agilisium used Google Cloud Platform (GCP) services to architect a commercial data lake and a data

Highlights

50%

of total operational cost at call centre

capacity planning to handle forecasted call volume’s

Increased productivity based on calculated metrics

Improved quality of services to customers

Improved performance

Client Profile

Vodafone, a global leader in telecommunications and technology services, is at the forefront of this transformation. Vodafone serves 625 million customers in 66 countries through its own and partner networks. The company's customer and network reach drive its mission to provide the technology and services necessary to build inclusive digital societies in its operating countries, while also halving its environmental footprint.

Business challenges

To migrate their on-premises system to GCP and to architect a commercial data lake and a data warehouse with end-to-end process automation includes web application to show the model output on dashboards to visualise the various KPI’s.
Enables business to unlock complex optimisations that leverage simulation to better understand the impact of choices on performance
Working Prototypes of Simulation showcasing the tactics and levers to be applied.
Intra-day optimising the application based on tactics to improve performance against KPI targets. For example, assessing the impact of postponing work (DOW) that does not require further caller interaction to improve utilisation at peak times
Tactics showing better shift planning for based on different levers.
Planning and assessing the impact of staffing decisions on likely operational performance and required headcount
Operations view showcasing the DOW and impact on KPI’s like average call wait time, abandon rate, call volume, AHT and service level.

Agilisium Solution

The solution design has three major components,

1. Data load and transformation

2. Simulation

3. Web Application

Both Front-end (Web application) and back-end (Simulation) are deployed on one GKE cluster segregated by namespaces.
The data from local markets are copied to the cloud storage location using Nifi service
Web application and backend services cluster responsible for rendering the dashboard
Simulator is used to run the data simulation and write output to CloudSQL and BigQuery
Dataflow is used to load the data from cloud storage to BigQuery and complete transformation
Cloud composer is used to trigger the dataflow jobs based on schedule
The web application is restricted for internal users.
The cloud armor is used to restrict the access and prevent any external traffic from reaching the web application. The internal users will have to be logged in before accessing the application. This is enforced through rules configured in Identity Aware Proxy.
Collibra for managing the data dictionary.

Tech-stacks Used

Ni-Fi

Google Cloud Storage

DataFlow

Composer-Airflow

VPC

Cloud SQL

BigQuery

Google Kubernetes Engine (GKE)

IAP

Cloud Armor

Load balancers

CMEK

Business outcome

Increase efficiencies in the in-call processing capabilities with existing staffing capabilities leading to cost savings.
Reporting and analysis on nearly real time basis for adherence to KPI’s for the call centre staff on Local Markets.
Showcase and apply multiple tactics, based on analytics and simulation on the historical data to apply changes on the real time basis.
Showcase the variance on KPI’s set for the call centre executive on nearly real time basis.
Multi-tenant GCP platform with dataset segregated for each market as a replica without affecting the existing operation