Tariff migration recommender: offer a tariff cheaper for the customer and better for the operator

Scenario

Customer X on the “Unlimited” tariff at 100k UZS per month. In fact uses 3GB of data and 200 minutes. On the “Medium” tariff at 60k UZS everything they use would still be included. The customer pays 40k more for a service they do not use.

A year later the customer will figure this out themselves, switch to a cheaper tariff or leave. From an ARPU standpoint it is “better to leave as is”. From a retention standpoint this is a churn bomb with a delayed fuse.

Recommender is the system that proactively suggests migration to the optimal tariff, before the customer does it themselves or leaves.

Recommendation logic

Per customer over a rolling 90 days analyse:

• Actual usage (minutes, GB, SMS).
• Current plan (limits, price).
• All available plans (catalog matching).
• Best-fit plan: minimum monthly cost that covers actual usage + 20% buffer.

If the best-fit plan is meaningfully (>15-20%) cheaper than the current — the customer is a migration candidate.

Reverse case also works: customer on a cheap tariff, regularly exceeds limits and pays overage. Best-fit may be a plan above basic but cheaper than total spend.

When to communicate

Not right after activation — the customer perceives it as “I only just signed up and you are already changing my tariff”.

After 6+ months on the current tariff. Enough data for a reliable recommendation.

After a significant change in usage pattern (for example, +50% data in a month) — a signal the customer entered a new phase.

During roaming, when extra-bundle spend appears, when a “too expensive” complaint is logged in support.

Action

Push with transparent wording: “Over the last 3 months you used on average 3GB and 200 minutes. On the ‘Medium’ tariff this is all included for 60k instead of your current 100k. Want to switch? [Compare] [Switch]”.

Compare — transparent side-by-side table. No hidden deductions.

Switch button — instant action. No office visit, no paper forms.

What is measured

Migration acceptance rate — what share of recommendations is accepted.

Customer satisfaction post-migration — usually grows sharply.

Churn rate post-migration vs control — the recommender must reduce churn, otherwise economics does not work.

ARPU impact — may decrease in the short term, but retention compensates.

LTV impact — the main metric, on a 12-18 month horizon.

What not to do

Do not recommend migration to a more expensive plan if usage does not justify. That destroys trust.

Do not migrate without explicit customer consent — even if better for both.

Do not ignore “honest fit” — a recommender optimised for ARPU degrades in trust quickly.

Do not push the recommendation every month.

How SamaraliSoft engages

Sprint Tariff Recommender Use Case — 6-8 weeks. Analysis of usage distribution, matching algorithm design, communication framework, pilot with measurement of acceptance / churn / LTV.

Related

• /en/insights/telecom-tariff-architecture/ — tariff architecture
• /en/solutions/telecom-cvm-platform/ — CVM
• /en/architecture/telecom-realtime-decisioning-architecture/ — decisioning
• /en/insights/telecom-customer-trust/ — trust