Every failed user-acquisition plan we have audited made the same substitution: it solved installs — a procurement problem with a price — and reported the result as users, which is a strategy problem with a retention curve. The gap between those two words is where UA budgets die.
Here is the strategy that survives contact with India’s market in 2026, in five decisions.
Decision 1: define the user before buying the install
Pick the early action that predicts lifetime value in your data — day-2 return, first transaction, level-5 completed, KYC done. That action, not the install, is what every channel gets optimised toward and, wherever reporting allows, priced on. Teams that skip this step optimise toward whatever each channel finds easiest to deliver, which is always the wrong thing.
Decision 2: build the mix in layers, not a list
- Algorithmic platforms — Google App Campaigns, Meta. The scale layer: strongest optimisation, least transparency, costs that drift with competition.
- Outcome-priced network — CPI moving to CPA. The economics layer: fixed unit costs, fraud screened before billing, sub-publisher control. This is the layer that makes the monthly plan a promise instead of a forecast.
- OEM stores — GetApps, Galaxy Store and peers. The cheap-reach layer for Android-heavy India; quality varies, so cap and cohort it.
- Creators — the trust layer for finance, health and anything requiring belief before download, tracked per creator or it is theatre.
Budget split that keeps discipline: roughly 70% proven, 20% scaling last cycle’s winners, 10% new bets — rebalanced monthly on cohort data, not opinions.
Decision 3: make retention the gate, not the report
Set your day-1 / day-7 / day-30 baseline from existing blended cohorts. Every paid source is then judged against it inside the MMP: within the band, scale; below it, cut — per sub-publisher, not per network average. Averages are where bad supply hides. A source delivering 60% of baseline D7 is not a cheaper source; it is a more expensive one wearing a smaller CPI.
Decision 4: treat fraud as a budget line you refuse to fund
India’s install-fraud economy is industrial: click injection, click flooding, device farms, SDK spoofing. The controls are mechanical — CTIT distribution monitoring, conversion-rate floors, device and datacentre-IP intelligence, post-install behavioural comparison — and the operational rule matters more than any tool: screening happens before billing. A partner who screens after complaints has made your finance team the fraud department.
Decision 5: plan the funnel migration in advance
Month one: CPI to establish volume and let sources reveal themselves. Month two: shift the billable event to registration once the MMP partner links are proven. Month three onward: CPA on the LTV-predicting action from Decision 1, with CPI kept only for channels that cannot support deeper events. Announce this path to partners on day one — the good ones optimise toward it early, and the ones who object have told you what they sell.
What the plan looks like on one page
| Layer | Pricing | Judge on | Kill rule |
|---|---|---|---|
| Google / Meta | Platform bidding | Cost per D7-retained user | CPA drift > 25% for 2 weeks |
| Network | CPI → CPA | Cohorts vs baseline, per sub-ID | Sub-ID below retention band |
| OEM | CPI, capped | Cohort quality | Below band after 2 reads |
| Creators | Flat + CPA | Cost per install per code | No conversions on test |
If you want the network layer of this plan priced for your app — installs, registrations or in-app events at a fixed cost with the fraud screening included — send the event and your geos. Quote same day, live inside 48 hours of the MMP link.
Written by the Performetra campaign team. If you want this applied to a live campaign rather than read about, tell us what you are running.