App Marketing in India: The User Acquisition Playbook

The same spend looks cheap or expensive depending on which box your dashboard stops at.

An install is a download, not a customer. An acquisition plan built around cost per install optimises for the one number in the dashboard that does not pay salaries.

This is the cross-category playbook: consumer apps, D2C shopping apps, marketplaces, services, subscriptions. The fintech version (KYC drop-off, first-transaction activation, lending ad policy) lives on our fintech app user acquisition guide and is deliberately not repeated here. For app-led SaaS, the wider motion sits inside SaaS digital marketing.

Why install count is the wrong north star

Install count is a volume metric any budget can move without telling you whether the product works. Replace it with the first in-app action that tracks revenue: a first order, a second session, a finished profile, a subscription started. Report cost per that.

• Pick the earliest event that predicts revenue, not the most impressive one. For a grocery app that is the first order placed, not "added to cart".
• Make it fire reliably before you spend. An activation event that misfires on a slice of Android builds teaches the bidding algorithm the wrong lesson for a month.
• Hold two numbers side by side: cost per activated user and day-30 retention. The first tells you what you paid; the second tells you whether it was worth paying. Our customer acquisition cost guide covers the broader CAC maths this sits inside.

What you actually control in an app install campaign

Less than most briefs assume. You supply assets, a bid, a budget, languages and locations; Google assembles the combinations and decides where they run. Google Ads Help's About App campaigns page (checked 23 September 2026; no update date shown) says the system tests asset combinations and serves the best-performing ads across Search, Google Play, YouTube, the Display Network and Discover.

• Google, assets: the App campaigns setup best-practices guide documents up to 10 text assets, 20 images, 20 videos and 20 HTML5 assets per ad group, and recommends videos of 10 to 30 seconds in portrait, landscape and square so they qualify across more channels.
• Google, bidding: About bidding in App campaigns lists target cost per install, maximise conversions for installs, target CPA and maximise conversions for in-app actions, target cost per pre-registration, and target ROAS. Target ROAS requires the Google Analytics for Firebase SDK, which makes it a build decision before a media one.
• Meta, events: Optimize Your App Ad lists the standard app events available for optimisation and says other standard events are not supported. Meta also directs advertisers to install the SDK rather than optimise for link clicks.
• Apple, placements: Apple Ads' Ad Placement Options documents four: Today tab, Search tab, search results and product pages, all charged cost-per-tap, with creative built from your App Store Connect metadata. Only search results takes keywords; Apple's Structure Campaigns guide recommends exact match for brand, category and competitor campaigns, broad match plus Search Match for discovery.

The acquisition mix, channel by channel

No single channel fills an app in India. The useful question is what each is good at, what it cannot tell you, and how it behaves in an Android-heavy market.

Take Thela, a hypothetical Pune grocery app, Android-first, ₹5 lakh a month. An honest mix: Google carrying volume, Meta carrying creative-led discovery, ASO fixed before either scales, referral built into the first-order flow, Apple Search Ads left out until iOS earns a line item. Building and running that mix is what our performance marketing team is for.

ASO is the only channel that keeps paying after you stop spending

App store optimisation decides how many people who reach your store page actually install. That makes it a multiplier on every paid click you already bought, not a separate organic project.

Google's Best practices for your store listing (Play Console Help, checked 23 September 2026) sets the hard limits: 30 characters for the app title, 80 for the short description, 4,000 for the full one. It also tells developers to use everyday language and avoid repetition of words. That is Google's way of saying that stuffing keywords into the description is not the tactic some ASO decks still sell.

The under-used feature is custom store listings. Play Console Help's Create custom store listings to target specific user segments documents up to 50 listing pages, targetable by country, by user state such as churned or lapsed or non-buyer, by pre-registration, by the Play search terms people used, and by Google Ads traffic. Each can change the app name, icon, descriptions and graphics.

So a Diwali offer campaign can finally land on a listing whose screenshots show that offer. The same page notes custom listings are not auto-translated. Each language is added by hand, which is why Hindi, Marathi and Tamil listings keep getting postponed.

Ratings are the one listing field you cannot write. Prompt for them when the app has just done something useful, not during onboarding.

Measurement after IDFA: what an MMP actually does

On iOS you no longer get user-level attribution. You get delayed, aggregated, privacy-thresholded postbacks. Apple's AdAttributionKit documentation (iOS 17.4 onward; checked 23 September 2026) sets the shape, and the numbers matter before you promise a live iOS dashboard.

• Three conversion windows: days 0–2, 3–7 and 8–35 after first launch, per Apple's Receiving postbacks in multiple conversion windows.
• Deliberate delay: the first postback arrives 24 to 48 hours after its window closes, the second and third 24 to 144 hours after theirs.
• Crowd anonymity tiers decide your detail. Tier 3 can return a fine conversion value and a country code; Tier 1 only a coarse value and a two-digit source identifier; Tier 0 one postback and nothing after. Low-volume campaigns get less data, by design.
• Meta's iOS limits are specific. Meta for Developers' Aggregated Event Measurement and Limitations of SKAdNetwork documents 9 campaigns across up to 9 ad accounts and 5 ad sets per Meta App ID on iOS 14+, 8 conversion events you can configure and prioritise, and no deferred deep linking for new iOS SKAdNetwork campaigns.

A mobile measurement partner stitches this together. Google Ads Help's About tracking app conversions with an App Attribution Partner names seven certified partners (Adjust, Airbridge, AppsFlyer, Branch, Kochava, Singular and Tenjin) and describes the job as attributing conversions, importing them into Google Ads and passing consent status. That is the honest scope: an MMP referees between platforms that each want to claim the same install. It does not recreate data privacy rules deleted.

India's split makes this less painful than it sounds. StatCounter Global Stats put Android at 92.79% of India's mobile OS market in August 2026 against 7.15% for iOS, so aggregated measurement governs a minority of an India-only app's installs. Build reporting on the Android truth, treat iOS as modelled, and plumb the events properly. Our conversion tracking guide covers that part.

The onboarding-to-activation gap where Indian app budgets leak

Our view, stated as opinion: the biggest waste we see in Indian app budgets is not bad targeting or weak creative. It is buying installs at the lowest possible cost while the first ninety seconds of the app destroys most of them, then scaling that campaign because the CPI looked good.

The arithmetic on Thela. These rates show the mechanism; they are not benchmarks, and there is no honest industry average to borrow.

The install-optimised campaign buys 2.4 times more installs and 53% fewer activated users. Every dashboard built on install count calls it the winner.

Closing the gap is mostly product work: onboarding that asks for a phone number and nothing else, a first screen showing real inventory before a signup wall, a visible language toggle rather than one buried in settings, and a download size a user on a metered connection does not resent. Heavy first-launch assets are paid for by your users, in data, before the app has earned anything.

Retention and re-engagement: the cheapest user is the one you already had

Re-engagement is a separate campaign type with its own economics, and most Indian app budgets turn it on too late. Google Ads Help's About bidding in App campaigns for engagement (checked 23 September 2026) covers target-CPA bidding for in-app actions by existing users, and advises a daily budget of 15 times your target CPA so the campaign has room to learn.

• Segment before you spend. Someone who installed and never opened needs a different message from someone who ordered twice and stopped; the logic is in our customer segmentation guide.
• Send lapsed users to a listing built for them. Custom store listings target churned and lapsed states, per Play Console Help above. Most apps never use this.
• Plan around the festive spike. Install intent rises on its own around Diwali and the sale weeks, and so does auction pressure; prep belongs in September, not the week of. Our festive season marketing guide covers the calendar.
• Deep link every push and ad. Dropping a re-engaged user on the home screen wastes the click.

Your first 90 days, and the gate at each step

The sequence matters more than the tactics. Each step has a gate: if you cannot clear it, fixing it beats moving on.

1. Weeks 1–2 — Define and instrument the activation event. Pick the earliest in-app action that predicts revenue and wire it into your MMP or Firebase. Gate: a test install fires it correctly on three device types. If not, do not launch.
2. Weeks 1–3 — Rebuild the store listing before you buy a click. Title inside 30 characters, a short description doing real work inside 80, screenshots that show the product rather than the brand, and a regional-language listing wherever you advertise in that language. Gate: store page conversion rate baselined.
3. Weeks 3–6 — Launch one Google App campaign and one Meta app promotion campaign on install bidding. Two campaigns, not eight. Gate: enough installs to read an activation rate you would defend in a meeting.
4. Weeks 6–9 — Switch bidding to the activation event. Google to target CPA on that event, Meta to the matching supported app event. Gate: the platforms report enough of it to optimise on. If volume starves, move the event one step earlier in the funnel rather than reverting to installs.
5. Weeks 8–10 — Decide on Apple Search Ads on evidence. Look at iOS share of activated users, not installs. Gate: iOS earns its own budget line, or it waits.
6. Weeks 9–13 — Turn on re-engagement and start ASO experiments. One lapsed-user campaign, one custom store listing per major segment, one listing variable at a time. Gate: day-7 and day-30 retention flat or improving before you scale. Scaling on leaking retention just buys the leak more volume.

Ninety days will not build a finished growth engine. It produces a number you trust. Our mobile marketing guide covers the surfaces around the app that this playbook does not.