Skip to content
Oliver Wyman, for a Gulf telecom operatorTelecommunicationsMiddle East

Oliver Wyman: Network Investment & 5G Optimization

Prioritize network upgrades by economic value, not technical severity alone.

Delivery roleConsultant, end-to-end analytics and modeling

Context

An Oliver Wyman consulting engagement in Dubai for a Gulf telecom operator deciding where to direct capacity and 5G investment across its network.

Constraint

A Gulf telecom operator needed to prioritize network upgrades across petabyte-scale network and customer data. Devices emitted a network ping every few minutes, creating a volume that made conventional analytics and model training difficult with the available compute. Technical degradation alone was not enough: leadership needed to know where a better network would create the most economic value.

Investment decision model
Where should the next network upgrade go?
Economic priority

The decision combines network-linked retention risk with customer value and intervention effort. Candidate labels are conceptual and preserve proprietary ranking logic.

System

Led the analytics and modeling work that joined distributed network-performance data with customer behavior and commercial value. Used Hive-based queries, partitioning, and aggregation to make event-level data usable at scale, then built a churn model relating network quality indicators such as effective speed and load to retention risk. Crossed that signal with customer value to create an investment-priority indicator for capacity and 5G interventions, without exposing the proprietary ranking logic.

Outcome

Reframed network investment as an economic prioritization decision: executive teams could compare upgrade scenarios by the customers affected, their retention risk, their value, and the expected return of an intervention rather than network severity alone.

How it was built

Approach

  1. 01
    Distributed analytics
    Combined device pings, network load, speed, congestion, geography, customer behavior, and commercial value at petabyte scale, using Hive partitioning and aggregation to make event-level data tractable.
  2. 02
    Churn modeling
    Built a binomial classification model relating network-performance variables such as effective speed and load to churn, estimating where improved conditions could plausibly improve retention.
  3. 03
    Economic bridge
    Crossed churn likelihood with customer value to bridge network engineering decisions and commercial impact, while keeping the proprietary ranking logic inside the engagement.
  4. 04
    Investment optimization
    Ranked candidate network interventions by expected return, weighing affected population, customer value, performance gap, and required investment, and translated the results into executive-level investment scenarios.
Related capabilities
  • Agentic Automation & MCP Integration - Agents that take multi-step operational work off your team end to end, built to the governance bar of enterprise and government environments: every action logged, auditable and reversible.
  • Data Platform & Cloud Foundations - Data your AI initiatives and analytics teams can trust: governed pipelines, a modeled warehouse and self-serve BI, built by someone who has shipped this in federal government and telecom environments.
  • Architecture & Implementation - From roadmap to production: I architect and personally build the systems the assessment prioritized, including production RAG, agentic workflows, evaluation frameworks, data platforms and the cloud beneath them, with governance and handover built in.
Direct booking, no form required

Discuss a similar initiative

30 minutes to map the decision, the data behind it, and the shortest credible path to a production outcome.

Book a 30-minute callor email info@yannickbounouar.com
Back to work ·View services·Contact