Compensation for poor network experience: four approaches and their real cost

Where the choice appears

In telecom operations there is a class of data showing that quality is poor for a specific subscriber right now. Not “an issue in this cell” but “this SIM has been dropping calls and seeing high jitter for the last hour”. This data is available in near-real-time through network probes, RAN counters and device telemetry — operators with mature analytics see them continuously.

What to do with that knowledge from the commercial side is not one scenario but four. Each has an applicability zone, a cost and a downside. Before picking one, it is important to admit what happens most often: nothing. The operator sees the data and does not act, leaving the customer to complain or quietly leave. That is the worst of the four.

Approach 1. Automatic compensation to everyone affected

The operator detects the incident, identifies the affected group and automatically credits compensation — bonus minutes, gigabytes, cash credit. The customer gets a message: “we noticed an issue in your area, we have added X to your account”.

When it fits. The incident is localised, the affected cohort is measurable, and the compensation has tangible value for the customer. For example, a multi-hour outage in a district — automatic credit to postpaid subscribers in that zone creates a strong feeling that the operator notices and reacts.

Where it becomes dangerous. If incidents are frequent and small, automatic compensation turns into background spend — the customer stops noticing it as a gesture and starts treating it as the norm. When an incident occurs and nothing is credited, complaints follow.

Cost is significant and non-linear. The more often automation triggers, the higher the margin load. And if the metric triggers on false positives, the operator compensates for incidents that did not happen — undermining trust in its own metrics.

Risk of gaming. Some customers learn to create conditions that trigger compensation — small problem on short outages, a larger problem on widespread ones.

Approach 2. Reactive compensation on complaint

The operator does no proactive work. If the customer calls and complains about quality, the contact centre has authority to grant compensation provided the data confirms the issue.

When it fits. Incident rate is relatively low and most customers tolerate small issues. Also when the operator wants to minimise compensation cost.

Where it breaks. The customer sees an issue and the operator stays silent. That creates a feeling of being unseen. The silent majority that does not call accumulates frustration and at some point leaves. By indirect signals, this is a meaningful churn driver, even though it is hard to measure directly.

Compensation distribution becomes unfair: those who call louder get more. On social networks this quickly becomes a reputational case “they gave it to one and not to another”.

Approach 3. Segmented proactive — compensation only for high-value

Proactive compensation for a specific high-value cohort: premium postpaid, corporate clients, long-tenure subscribers. Other segments are served reactively.

When it fits. The base has skewed LTV — a small share of customers contributes a meaningful share of revenue. Defending their retention covers proactive compensation, because preventing the churn of one large customer covers a meaningful part of the programme.

Where the mine is buried. If LTV is more evenly distributed, picking out the “premium” becomes arbitrary, and borderline customers feel offended. Lower-tier customers find out (through social networks, through friends) that the premium gets more for the same incident. This becomes perceived unfairness.

Approach 4. No money, but open communication

The operator does not give financial compensation but proactively reports: “we noticed N minutes of poor quality for you today, the incident lasted until X, we are working on the fix”. No financial gesture — only confirmation and context.

When it fits. The customer base is in principle tolerant of network issues and does not expect compensation. The operator can publish honest improvement statistics (not just dashboards for the press office) and back words with action. The brand is strong and credibility is high.

Where it breaks. In a competitive environment where neighbouring operators give financial compensation, customer tolerance falls. Acknowledgement without a financial gesture starts to sound like empty corporate words, especially if improvements are not visible to the subscriber.

Comparison on the things that matter

Criterion	Auto to all	Reactive	Segmented	Acknowledgement
Customer satisfaction	high initially	mixed	high for premium	low-medium
Direct cost	high	medium	medium	low
Main risk	gesture inflation	unfair distribution	tier resentment	empty words
Time to launch	6-9 months	1-3 months	9-12 months	1-2 months
Data requirements	high	medium	high	low
Sustainability over time	hard	good	good if segmented	depends on honesty

When to choose what

Auto to all — a strategic positioning move. “We compensate when you suffer” becomes part of the brand. It needs sustained investment and operational maturity. Doing it half-heartedly is worse than not doing it.

Reactive — a reasonable entry when there is no incident-rate baseline yet and the operator does not want to invest in infrastructure before getting signal. Can later evolve into proactive.

Segmented proactive — commercial common sense when LTV is skewed. Defend those whose churn costs the most, run reactive for everyone else. A good option if the organisation can explain the difference without offending the mass base.

Acknowledgement only — works in mature contexts where the customer values honesty over money. It needs a strong brand and visible real improvements. On a young competitive market it rarely lands.

What is often done incorrectly

A mix of approaches without a rule. Sometimes proactive, sometimes reactive, sometimes nothing. This creates a feeling of chaos and makes it impossible for the contact centre to explain the logic.

Compensation in a form the customer does not value. Giving 500 MB to a customer on an unlimited plan — a gesture exists, value is zero. The compensation has to be in a form relevant to the recipient.

Acknowledgement without operational improvement. Six months of “we know you have an issue” without resolution turns the messages into anti-advertising.

Compensation triggered by false positive. The network reports an incident that did not occur — the customer gets a credit for something they did not notice. This undermines trust in the company’s own metric.

Discussion points for the committee

What is the current incident detection capability — how granularly and how live can we see quality at a specific subscriber? The answer determines which approaches are technically realistic.

What does customer satisfaction show after network incidents today? If the survey shows nothing, measurement is needed before choosing an approach.

What is the LTV distribution in the base? That determines whether the segmented model is justified.

Does billing and the app support proactive notifications at the moment of an incident? Without that, the automatic approach cannot launch.

What are the closest competitors doing? Minimum baseline in a competitive environment is set by them, and ignoring it is unwise.

How SamaraliSoft can help

Network Compensation Strategy & Operating Model — diagnostic of the current incident detection capability, choice of approach against the structure of the base and LTV, design of the operating model for the chosen approach (roles, triggers, limits, escalation), integration plan with network ops and the contact centre, and a 90-120 day pilot with metrics that say go or no.

Related reading

• /en/insights/telecom-network-experience/ — network experience as a commercial asset
• /en/use-cases/telecom-churn-war-room-mnp/ — retention in the MNP era
• /en/insights/telecom-subscriber-intelligence-operating-model/ — operating model
• /en/insights/telecom-nba/ — NBA as context

Sources

• Ministry of Digital Technologies — coverage and speeds UZ
• Digital Government Portal — internet usage
• UzDaily — telecom market
• MNP.UZ — number portability