The Operator's Playbook: How to Build Products That Win in Indian Telecom

Why Indian Telecom Is Unlike Any Other Market

In most markets, a product manager's job is to find the right segment and serve it well. In Indian telecom, you are simultaneously serving a corporate CFO in Mumbai on a 5G postpaid plan and a daily-wage earner in rural Rajasthan on a 28-day prepaid recharge worth INR 179. The same network. The same billing system. The same P&L.

This is not just a segmentation challenge. It is a resource allocation problem at civilisational scale. Every pricing decision, every network investment, every product feature you launch ripples across 300-500 million users with vastly different contexts, devices, literacy levels, and relationships with data.

Add to this: spectrum is a finite, government-allocated resource that costs billions. Network economics mean your margins in a rural circle are structurally weaker than urban even if ARPU is similar. And after the Jio disruption of 2016, the entire market repriced overnight - ARPU collapsed from INR 170+ to sub-INR 100, and operators had to rebuild their entire product logic on a new economic floor.

Most global playbooks break here. The three-layer model is what actually works.

The Three-Layer Model

Think of any telecom product decision as having three layers, each with its own constraints and levers. The mistake most PMs make is trying to solve a Layer 3 problem with Layer 1 thinking, or vice versa.

Layer 1 - Network Reality. This is your foundation. Spectrum positions, coverage depth, technology generation (3G/4G/5G rollout status), and quality of service metrics by geography. You cannot build a product promise your network cannot keep. A 4G video streaming plan in a circle where your 4G coverage is 40% is a churn machine, not a revenue driver.

Every product decision must start with a Layer 1 audit: does the network support this? What percentage of my target segment actually lives in covered areas? What is the real throughput they will experience, not the theoretical maximum? This layer is largely not in your control as a PM - but it sets your ceiling.

Layer 2 - Market Segmentation. Indian telecom has the most complex segmentation matrix of any consumer market I have encountered. The primary cuts are urban vs rural (not just income - usage patterns, device capability, recharge behaviour, and churn triggers are fundamentally different), prepaid vs postpaid, and ARPU band (sub-100, 100-200, 200-500, 500+).

But within these, you have secondary segments that matter enormously: the migrant worker who recharges irregularly but never churns because of incoming calls from family; the small business owner who uses WhatsApp Business and consumes 10x data but will not pay postpaid rates; the student who is price-sensitive but highly social and drives viral adoption. Your product architecture needs to speak to each of these differently.

Layer 3 - Revenue Architecture. This is where the actual product work happens. Base plans, add-ons, bundle mechanics, validity structures, loyalty triggers, and retention levers. The key insight: in a low-ARPU market, value perception matters more than absolute price. A plan at INR 299 with 3GB/day feels different from a plan at INR 299 with unlimited data and throttling at 3GB. Same economics, very different conversion.

The second insight: validity is a product variable, not just a billing parameter. The 28-day vs 30-day vs 56-day vs 84-day recharge cycle is a churn management tool. Understanding when your segments run out of balance - and ensuring your plan validity lands them back at recharge decision on your preferred day - is a legitimate product strategy.

The ARPU Paradox

India has some of the cheapest data in the world - under $0.10 per GB in many plans. Yet the operators running these networks are building some of the most sophisticated AI infrastructure on the planet. How?

The resolution is scale. At 300-400 million subscribers, even a 1% improvement in churn is 3-4 million users. A product feature that lifts ARPU by INR 10 per month generates INR 3-4 billion in annual revenue. The economics are unit-level thin but aggregate-level massive.

This means precision matters more than in any other market. You cannot afford broad, expensive moves. A targeted retention intervention for a specific microsegment - identified by AI, triggered by a specific behavioural pattern, delivered through the right channel at the right moment - will outperform a blanket offer every time. The operators who understood this early built data science teams before it was fashionable.

AI at Scale Means Something Different Here

AI in Indian telecom is not a nice-to-have. At 300M+ subscribers, the data volume is extraordinary. Call records, data usage patterns, location data, recharge history, customer service interactions - the dataset for a single operator is one of the richest consumer datasets in the world.

The product applications are significant: churn prediction models that identify at-risk subscribers 30 days before they port out, with enough lead time to intervene. Dynamic pricing engines that adjust offer recommendations in real time based on individual usage patterns. Network load prediction that lets you pre-position capacity before demand spikes rather than reacting after.

But here is the important nuance: AI is a Layer 1 and Layer 2 tool in telecom, not primarily a Layer 3 tool. It helps you understand your network and your segments with far greater precision. The Layer 3 product decisions still require human judgement - because they involve regulatory constraints, competitive positioning, and brand considerations that models do not naturally weight correctly. The PM's job is to translate model outputs into product decisions, and to know when to override the model because it is optimising for the wrong thing.

Three Rules That Actually Work

• Never price against the network. A product promise your network cannot deliver will generate more churn than the revenue it creates. Honest products - even if competitively inferior - build more durable customer relationships than oversold ones.
• Segment by behaviour, not by demographics. Age and income are weak predictors of telecom behaviour. Usage pattern, recharge regularity, and channel preference are far stronger. Build your segments around these, and your products will land more precisely.
• The retention offer is already too late. By the time a subscriber is calling your retention desk, you have already lost the value battle. The best retention happens at the product design stage - build plans that create habitual usage, not just transactional consumption.

Vinay Mangal is Head of Products at Ooredoo Maldives. He has spent 25 years building telecom products across Idea Cellular, Vodafone Idea, Reliance Communications, and other operators. He holds a Six Sigma Black Belt and has managed P&Ls spanning 300M+ subscribers.