Network experience as a commercial asset, not an engineering metric

A number that explains the market

Between 2020 and 2025 the weekly data volume at one operator in Uzbekistan rose from 2,000 TB to 23,600+ TB. The 4G traffic share moved from 57.7% to 93.4% (Beeline Uzbekistan). National mobile data speed reached 51.42 Mbit/s (dgov.uz).

This number is not about infrastructure. It is about consumer behaviour. The customer today uses data on an order more intensively than five years ago. Every hour of quality drop is not a “technical incident” — it is a degradation of experience in the channel the customer spends hours a day with. And it directly affects churn, ARPU and support cost.

In spite of this, network experience in most operators stays an engineering topic. Engineers’ KPI is uptime, capacity, packet loss. Commercial KPI is ARPU, churn, sales. The link between these two worlds is weak or missing. That creates three problems visible in the P&L.

Three places where the gap costs money

First — reactive customer care. The customer calls support after a poor connection experience. A typical first-contact resolution rate is 60-70%, meaning 30-40% of cases need a second contact, escalation, an engineer dispatch. This is operational load that costs every operator a substantial annual figure. If the network problem were detected before the customer called and a proactive notification were possible, support cost drops by 15-25%.

Second — silent churn. The customer does not call, does not write, does not file a complaint. They simply move to another operator. With MNP this became faster and cheaper. Indirect signals — a sharp 50-70% traffic drop over a week in a zone with stable coverage — can predict a meaningful share of these departures. Without integrating network data into retention processes, the operator finds out about the departure 30 days later, when it is too late.

Third — inefficient network investment. CapEx is allocated against a coverage plan, not against impact on retention and ARPU. As a result the operator over-invests in zones with surplus capacity and under-invests in zones where moderate degradation drives premium-segment churn. That is a planning gap.

What “network experience as a commercial asset” means

It is an operational and architectural link across four layers.

Layer 1: collection of QoE signals. Data from the network on signal quality per cell, packet loss, latency, throughput, dropped calls. This is already collected at any modern operator. The question is where this data flows.

Layer 2: customer-level attribution. Linking network events to specific customers rather than cell-level aggregates. If cell 12 had a 4-hour issue, which specific customers used data during that time and experienced degradation? This requires real-time merging of network logs with CRM.

Layer 3: business action engine. When a high-value customer experiences meaningful degradation, what should happen? Options — proactive notification (“we had technical issues, recovered”), compensation (bonus minutes or GB), inbound call with extra service, deferred upgrade offer. These actions must be pre-defined and automated.

Layer 4: feedback loop into network planning. Data on where degradation correlates with churn must reach the team that plans CapEx. “In zone X, 15% of premium customers churned this quarter, and average latency in this zone was 30% worse than the chip” — this changes investment priorities.

Without all four layers, network data stays a technical artifact. With them, it becomes a commercial asset.

Architecture of the layer

Technically the layer is built from existing components and does not require a full network re-engineering.

The QoE collector is standard OSS functionality at most operators. If missing — added through CDR (call detail records) and probe data from radio access network. Granularity matters — preferably per-cell, per-hour minimum, ideally per-cell per-15-min.

The customer attribution layer links CDR logs to customer ID through a series of joins by SIM, IMEI, location, timestamp. New component in most operating environments. The complexity is privacy: linking customer to network performance creates data sensitivity, requires consent management and audit log.

The business rules engine decides when a network event triggers commercial action. Rules: which level of degradation (latency >X, packet loss >Y%, downtime >Z minutes) for which customer segment triggers which action. These rules must be parameterisable and revised regularly.

Action delivery — proactive notification via push/SMS, compensation generated in the biller, escalation flag in CRM for the contact centre. Each action must be authorised and audited.

Network planning feedback — pipeline from retention/churn data back into RAN planning. This is organisational, not technical. Requires regular meetings between commercial and network leadership with a shared dashboard.

Concrete actions that work

Proactive compensation. If a cell zone had an incident for 4 hours during peak time, customers who used data during that window receive 1-3 GB or 30-60 minutes automatically. Notification: “We detected technical issues in your area and added a bonus package as compensation”. Cost to the operator is low (often incremental cost is near zero), the NPS effect is meaningful, churn declines.

Pre-emptive outreach to high-value customers. If degradation occurs in a zone with above-median ARPU customers, the contact centre calls proactively. “Wanted to let you know we had technical issues, the situation is restored. If there are questions — call us, we will add a bonus as a thank-you for your patience”. High-touch and expensive, but for the top-1% it pays back.

Network status in the app. A real screen in the telco app showing “all good in your area” or “issues detected, working on it”. This puts the operator ahead of customer complaints. Many issues are resolved by simply informing — the customer sees the operator knows and is acting.

Churn early warning. A sharp traffic drop for a customer in a zone without mass issues — an indicator of switching to another operator or preparing to port. The customer enters the retention queue with priority, with action being a call within 24 hours and an understanding of the issue.

Heat maps for CapEx. A map overlaying premium-customer density and quality KPI. Where there are red spots — that is where CapEx goes. Changes planning priorities from “uniform coverage” to “protect margin”.

What almost never works

Mass coverage map publication. Detailed coverage maps with colour zones rarely produce a commercial effect. Customers do not choose an operator from a map; they choose by recommendation and friend experience. Coverage maps are useful for the regulator and for self-image, not for retention.

In-app self-service troubleshooting. The idea of “let the customer test the network” is attractive in theory; in practice customers do not use such features. Those who call will call; those who quietly leave will not open self-service.

Stand-alone NPS surveys. Calling customers with network quality questions disconnected from retention processes produces metrics but not actions. NPS works when tied to operating model: low NPS triggers retention action, high NPS triggers loyalty offer.

Generic “quality improvement” without targeted attribution. “We are improving the network” is a message that does not work. Specific “we resolved the issue you experienced last Monday at 19:00” — works.

When not to launch

If the link between network and commerce is organisationally complex (network reports to CTO, commerce to CCO, and they meet only at C-level), launching the project requires CEO sponsorship. Without it the initiative gets stuck at the first real issue when a process change is needed in one of the teams.

If QoE data is not collected at per-cell per-hour granularity, or if collected but stored in a format hard to join with CRM, the project will hit data engineering at the first stage. This is solvable but requires 6-9 months of preparatory work.

If consent management does not cover the use of network data for personalisation and retention, the project creates regulatory risk. The personal data law requires targeted consent.

If the operations team is not ready to support automated actions (compensations, notifications), the system will fix bugs continuously in the first months. Not a reason not to launch, but a reason to allocate resources.

Discussion points for the committee

What is the cost of a support contact at your call centre today? How many such contacts per month? What share is connected to network experience? With these three numbers a first business-case estimate appears.

How many premium customers did you lose last year, and what share left without complaints (silent churn)? That is the second area where network experience can deliver impact.

What share of your network CapEx is based on coverage (uniformity) versus ROI (margin protection)? If the answer is “100% on coverage” — you are leaving money on the table.

How SamaraliSoft can help

Network Experience & QoE Intelligence — a diagnostic of existing network data flows, design of the customer attribution layer without changes to OSS, selection of 2-3 priority actions with a 90-day pilot plan, design of the feedback loop into CapEx planning. Not “we will sell you a new QoE tool” — you most likely already have one. The task is to connect it with retention and P&L.

Related reading

• /en/use-cases/telecom-churn-war-room-mnp/ — retention in the MNP era
• /en/insights/telecom-subscriber-intelligence-operating-model/ — operating model around data
• /en/insights/telecom-growth-after-connectivity/ — where the money sits beyond connectivity
• /en/architecture/telecom-around-core-architecture/ — growth layer around the core

Sources

• Digital Government Portal — 51.42 Mbps mobile, 91.79 Mbps fixed
• Beeline Uzbekistan — traffic growth 2020-2025
• Ministry of Digital Technologies — 98% coverage
• UzDaily — telecom market 24,873 billion UZS