The India landscape
India's paid app ecosystem has substantial Android usage, familiar instant-payment behaviour, price sensitivity and multiple material language markets. Google Ads, Meta and Apple Ads are the spine; ShareChat, Moj, Glance, InMobi and Moloco are expansion channels once that spine is instrumented.
Unit economics that survive scale
Work backwards from allowable CAC to allowable CPI. Allowable CAC = ARPU × payback horizon × margin. Allowable CPI = allowable CAC × install-to-payer conversion rate. Publish the number in the App Growth Truth File; every bid, budget and channel decision must reconcile against it. Use a clearly labelled scenario to test the formula with the app's own ARPU, payback, margin and install-to-payer rate. Any campaign above that calculated ceiling requires an incrementality justification.
The event and signal ladder
Install is the weakest signal you can optimise against. Climb the ladder: install → open → key-activation event → trial-start → subscribe/purchase → D7 retained-payer. Every rung up narrows volume and improves quality, up to the point the platform's learning phase starves. Use observed event volume to decide whether the platform has enough signal at the current rung. If it does not, optimise to the next reliable event.
Launch readiness and the first 72 hours
Before spend goes live: configure the MMP, test deep links end to end, document consent and DPDP posture, map iOS conversion values and reconcile product analytics to the MMP within an agreed variance. During the initial learning window, monitor before making bid, creative or structural changes.
Google App campaigns in India
Three campaign types, one decision tree. Install-volume for cold-start category signal. In-app-action for scaled apps with reliable post-install events. Value/ROAS for subscription and commerce apps with revenue-eventing at scale. Choose campaign structure by objective, operating system and market, with enough distinct assets to test format and message without creating clutter. Weekly decision order: budget check → tCPA/tROAS integrity check → asset rotation → geo/device audit → then and only then, structural changes.
Meta on Android vs iOS
Meta Android acquires cleanly against install and app-event optimisation. Meta iOS post-ATT requires disciplined use of Aggregated Event Measurement, a documented attribution window and tolerance for reporting lag. Do not read iOS Meta results daily; you will over-react to noise. Audience modes: broad + advantage+ audience for scale; interest-stacked for cold-start; retargeting only when the CRM or first-party audience has enough matched users for stable delivery.
Apple Search Ads in India
India is a lower-cost Apple Ads market than US/UK with meaningful conversion when metadata is engineered for capability terms in Indian-English. Structure discovery, brand, competitor and category intent separately, then use negative keywords to prevent overlap. Harvest search terms and prune on a regular cadence. CPP/CSL routing: every campaign points to an intent-matched Custom Product Page.
Regional-language creative
English-to-Hindi transcreation is not translation. Test transliterated Hindi against Devanagari rather than assuming one form wins. Give Tamil, Telugu, Bengali and Marathi standalone creative sets when observed audience value and volume justify them. Regional creators can outperform headquarters-produced UGC when language and context fit the audience. Prioritisation score: (market install share) × (category ARPU index) × (creator supply score).
Channel expansion gate (ShareChat, Moj, Glance, Moloco)
The gate: spine channels (Google, Meta, Apple) must be at scale with CAC inside allowable before any expansion channel gets a test budget. ShareChat and Moj deliver reach in tier-2/3 and vernacular audiences; Glance is lockscreen and works for utility and content apps; InMobi is programmatic display with strong DSP integration; Moloco is machine-learning DSP that shines on rich event signals. Set the expansion test budget from allowable CPI and a pre-agreed sample requirement; otherwise the signal may be too weak to interpret.
The weekly operating loop
Monday: read the previous week's cohort against benchmark. Tuesday: budget and bid changes (one variable per campaign per week). Wednesday: creative ship (winners propagate, losers archive). Thursday: audience/geo/device audit. Friday: reconciliation across MMP, Firebase, product analytics. Every finance number tied to a source. Every claim tied to a measurement. Every 'we grew' backed by cohort math, not a screenshot.