Retention economics: when retaining a customer costs more than letting them go

Simple arithmetic

Take an abstract customer. Monthly ARPU 50,000 UZS, service margin 30% — that is 15,000 UZS margin per month. The customer signals intent to leave, the retention team offers a save: 25% off for 6 months plus a bonus data package costing 30,000 UZS at cost.

Calculate the economics of retention. The discount eats 12,500 UZS of margin per month for 6 months — 75,000 UZS of foregone margin. Plus the bonus package at 30,000 UZS cost. Total retention investment — 105,000 UZS.

What we get in return. If the customer stays 12 months after the offer, average monthly margin is 15,000 minus 12,500 for the first 6 months — 2,500 UZS per month for half a year and 15,000 for the next half. Total margin over 12 months — 105,000 UZS.

Net effect — zero. We spent 105,000 and got 105,000 back over 12 months. No profit, no ROI on the retention investment.

This is not a theoretical example. It is the typical picture under retention offers without discipline. A large share of retention actions have this kind of structure and do not bring value above breakeven. Worse — many do not even reach breakeven, because the customer leaves before the discount ends or because actual margin is below the model.

Now take another customer. ARPU 20,000, margin 6,000 a month. The same offer with 25% off and a bonus is 99,000 UZS investment against 18,000 UZS margin × 6 months — 108,000 UZS over half a year (12 for the remaining half). On this segment retention slips into the negative.

What follows from this

Retention should not be done automatically for every customer who signals leaving. Retention has a boundary, and the boundary depends on segment, ARPU, actual margin, retention rate, expected lifetime after retention.

Many operators know this in theory but in practice the retention team works under a simple rule: “the customer is calling about leaving — give everything in the script as long as they stay”. The team’s metric is retained subscriber count, not retained margin.

This produces a situation where the operator spends millions on retaining customers with negative unit economics, with no separate visibility in the P&L.

Four scenarios for clarity

A well-built retention economics framework usually carves out four zones.

Zone 1. High-value customer. ARPU well above average, low service cost, long tenure. Worth retaining within reasonable bounds — LTV is high, retention investment pays back many times. The retention team’s focus.

Zone 2. Mid-value steady customer. Average and stable ARPU. Retention makes sense if the cost is not above 2-3 monthly margins — there is return, but no premium. Careful offers, not “give everything”.

Zone 3. Low-value but active customer. ARPU is low but the customer uses the service. Retention only makes sense at minimal cost — a simple message or a small bonus. A noticeable retention investment does not pay back.

Zone 4. Loss-making segment. The customer has negative margin (uses subsidised services, frequent complaints, abuse pattern). Retention here goes negative for the operator. Their departure improves economics.

The boundary between zones should not be on intuition — a real LTV model is needed, refreshed regularly. Without it the retention team works on instinct, and instinct almost always says “save them at any cost”.

What is often done incorrectly

A retention metric measured as count, not margin. The team’s KPI is “retained X customers this month”. This creates a perverse incentive — better to retain 100 low-value than miss 50 high-value.

A one-size offer without segmentation. One script, one discount for everyone. The most common and least effective approach. In high-value segments the offer is too small; in low-value it is too generous.

Permission to “add” to the retention offer without economic check. The agent sees the customer hesitating, adds another bonus, another price reduction, without checking whether it stays inside the cap. No hard cap on retention investment by segment.

No measurement of actual outcome. The customer stayed — victory. What happens 6 or 12 months later — not tracked by the retention team. So learning is hard — there is no feedback on which tactics work.

Retention when churn is driven by network or service. If the customer leaves because of poor coverage in their area or because of an inconvenient service, a retention offer does not solve it. They get a discount, stay for 3 months, leave anyway. This needs to be distinguished.

What a working retention model needs

LTV model by segment. Not one number — a distribution by micro-segments, refreshed.

Hard cap on retention investment. For each segment the maximum offer size that has ROI. The agent cannot exceed it.

Segment-level playbook. Different offers for different segments. High-value has access to additional tools (premium support upgrade, early access to new services). Low-value has a minimal list of simple offers.

Reason code for departure. Each retention case is tagged with a reason. If the cause is network — it is rerouted to network ops for fix, not “closed” with a discount. If competitive offer — that is a retention case. Financial reason — separate playbook.

Measurement at 6 and 12 months. What happened to the retained customer. Which tactics produced real return, which did not. Feedback loop for improvement.

When not to launch a retention reform

If the organisation does not have data on actual margin at customer level, an LTV model cannot be built. Data foundation first.

If the retention team is outsourced with rigid retained-count KPIs, reform hits the contract. Without renegotiation the reform does not start.

If sales and retention metrics are not separated — selling to a retained customer is recorded as a new sale — the incentive is distorted. This has to be fixed first.

If the CFO is not ready for real visibility into retention investment in the P&L (which will reveal losses that were previously hidden), the reform is politically blocked.

If the brand is positioned as “we retain at any cost”, a public rebrand is required. Without that, segmented retention is read as worsening for some in favour of others.

Discussion points for the committee

What is the current retention investment as a separate P&L line? If unknown or merged into general marketing — that is the first gap.

What is the LTV distribution by segment? And how often is this model refreshed?

What KPIs does the retention team have today? If only count — a rebuild to margin is needed.

What are the 5 most common reason codes for departure? That is the direction for fixing root causes proactively, rather than reactive retention.

Is the board ready for a temporary reduction in retention rate in exchange for improved retention margin? Without that commit the reform will not pass.

How SamaraliSoft can help

Retention Economics Diagnostic — analysis of actual retention investment unit economics by segment, build of a simple LTV model, segment playbook with hard caps on offers, measurement framework at 6/12 months, and a 90-day pilot reform on one segment.

Related reading

• /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-arpu-bundles-devices/ — where ARPU grows
• /en/insights/telecom-nba/ — NBA as context

Sources

• Digital Government Portal — digital infrastructure UZ
• UzDaily — Central Asia telecom market
• Ministry of Digital Technologies — telecom in UZ
• MNP.UZ — number portability