How to Promote an App After the Product Is Ready to Retain Users

App marketing is everything you do to get the right people to install your app, reach its value and keep coming back: positioning, pre-launch, store pages, channels, onboarding, retention and measurement. It is one funnel, not a list of tactics, and half of that funnel lives inside your product. Buy traffic for an app that can’t deliver its promise and you haven’t marketed it. You’ve paid strangers to find out it doesn’t work.

One common growth problem starts at the end. “We launched, we ran ads, we got installs, and nobody stayed.” The ads worked. The installs were real. The money was real too. What failed was everything after the install: a store page promising one thing, an onboarding asking for six things, and a home screen that looked nothing like the ad. Marketing brought people to the door. The door opened into a wall.

This guide treats app marketing the way it actually behaves: as a decision system that runs from the first time a stranger hears about your app to the day they open it without a reminder. It covers who the app is for and what promise to make, what must be ready before you promote anything, pre-launch and beta programs, app store conversion, how to choose a small channel mix instead of all of them, deep links and onboarding, retention, the metrics that belong to each stage, and when to scale, change or stop a channel.

It will not give you a list of 47 growth hacks. You can’t afford 47 channels, and the three that matter are easier to find once you stop looking at the other 44.

What app marketing is, and what it isn’t

Mobile app marketing covers the whole lifecycle: awareness (people learn the app exists), acquisition (they reach the store and install), activation (they get the core value for the first time), retention (they come back) and monetization (they pay, subscribe or generate revenue). App promotion is the loud part at the front. App advertising is one paid slice of promotion. Treat either one as the whole job and you are measuring the invitations, not the party.

Here is the uncomfortable part for anyone who thinks marketing ends at the install. Three of those five stages happen inside the product. Activation is onboarding and the first useful screen. Retention is whether the app is worth opening again. Monetization is whether the value is clear enough to pay for. A marketing team can fill the top of the funnel beautifully and still watch the money drain out through a product nobody designed for the people it just paid to bring in.

It is sending five hundred invitations to a house party where nobody has bought food. The invitations were lovely. The turnout was great. Everyone left in ten minutes, and now they tell their friends your parties are rubbish.

So when we say app marketing in this guide, we mean the full chain. Each channel reaches a person in a different context, makes a promise, and sends them somewhere. Whether that promise is kept is a product question. If you only remember one sentence, make it this one: marketing and product experience are one funnel, and you pay for every leak in it twice, once to acquire the user and again to replace them.

Find where your funnel leaks before you pay for more traffic

Most teams that ask us how to promote an app already have some traffic. What they don’t have is a clear view of where it disappears. Before you pick a channel, find the leak. More water in a leaking pipe doesn’t make the tap run faster. It makes the floor wetter.

This is the first pass we run when a team brings us an app with growth problems. Each symptom points to a stage, a likely cause and the people who own the fix.

Look at the right-hand column. Most of the rows are not “buy more ads”. That is not modesty on our part. It is the pattern we see in our app audits: acquisition gets the dashboard and the invoice while the journey after the install keeps losing the people you paid for. Fix that journey before paying to fill it again.

Decide who the app is for and what it promises

Every channel, every ad and every store screenshot makes a promise. If the team hasn’t agreed on what that promise is, each of them improvises, and the user meets five slightly different apps before reaching the real one.

Positioning answers three questions in plain words: who is this for, what problem does it solve for them, and why should they believe this app solves it better than what they do now. “The all-in-one productivity app for everyone” answers none of them. “Shift swaps for restaurant staff, agreed in two taps, without the group chat” answers all three. The second one is narrower. It is also the only one a stranger can remember after scrolling past it.

Narrow positioning scares founders because it seems to shrink the market. It doesn’t. It shrinks the part of the market you pay to reach and fail to convert. A promise aimed at everyone is a billboard in a language nobody reads: seen by millions, understood by none, invoiced in full.

Where does the promise come from? Not from the pitch deck. From evidence about the people you want: what they do today, what they complain about, which words they use, and what they’d stop doing if your app worked. If you don’t have that evidence yet, get it before the first campaign. A handful of interviews and a look at how people describe the problem in their own words costs less than one week of paid traffic aimed at the wrong promise. When a team has no validated audience needs and no reason to prefer one channel over another, that is the moment for proper UX research on your audience and positioning, not for a bigger ad budget.

A good positioning statement also tells you what not to promise. If your app is great at shift swaps and mediocre at payroll, the payroll feature stays out of the ads. Every promise the product can’t keep in the first session becomes a one-star review with your cost per install attached.

Define activation, not installs

An install is not a customer. It is a person who tapped a button in a store and hasn’t yet discovered whether your app is any use. Counting installs as success is counting how many people walked into the shop and ignoring whether anyone bought anything.

Activation is the first moment a new user gets the core value your app promises: the first shift swapped, the first budget set up and synced, the first workout logged with a plan for tomorrow. It is a specific event you can track, not a vibe. Pick it before launch, because every channel decision later depends on it. A channel that brings cheap installs who never activate is not cheap. It is expensive in a way the invoice doesn’t show.

This is Goodhart’s law with an app icon: when installs become the target, they stop measuring anything useful. Teams optimise for cost per install, campaigns learn to find people who install and leave, and the dashboard looks healthier every week while revenue stays flat.

A useful activation event has three properties. It happens early, ideally in the first session. It reflects the core value, not an admin step like “created an account”. And it predicts retention: people who reach it come back noticeably more often than people who don’t. You won’t know the third part for sure until you have data, so start with your best hypothesis and check it after the first few weeks of real users.

Write it down in one line and share it with everyone who touches acquisition: “A user is activated when they have swapped their first shift.” From that moment, every channel is judged on cost per activated user, not cost per install. The conversation about which channels work changes overnight.

What must be ready before promotion starts

Promotion is an amplifier. It makes whatever you have louder, including the broken parts. So before any money goes into reach, check that the thing being amplified can survive the attention.

The readiness gates we use are short, and every one of them is about the path a new user will actually walk:

• The promise is written down. One audience, one problem, one promise, agreed across the team.
• The store page matches the promise. Title, subtitle, screenshots and the first lines of the description say the same thing as the ads will.
• Onboarding reaches value in the first session. No six-screen tour, no ten-field profile before the user has seen why the app exists.
• Activation is defined and tracked. You can see who reached the core value, from which channel, in which cohort.
• Links land where they promise. Campaign links open the right screen in the app, or the right store page when the app isn’t installed.
• The core journey works on real devices. Not in the demo build. On the old phone with the small screen and the flaky connection.
• Someone reads the reviews and support messages. Early users tell you what is broken for free. Ignoring them is paying for the same lesson later.

If you are planning the product and its route to market at the same time, these gates are not a marketing checklist. They are part of the build. The deep link handling, the analytics events, the onboarding path and the store assets all need engineering and design time, and they need it before launch week, not during. That is the difference between a product that is technically finished and one that is ready to launch as a mobile product.

App launch operations: who owns the release day

A channel plan is not a release checklist. Before publishing an announcement, assign a named owner to each handoff and test the support route alongside the product.

Choose the announcement time around actual availability and support coverage. A release delayed by store review should not leave an already-sent campaign pointing at a promise nobody can use. Prepare reusable real screenshots and a concise product explanation for partners or press; distribution is useful only when the recipient has a relevant audience and an accurate destination.

A real product-to-website handoff: Runners High

Runners High combined an iPhone music experience with a marketing website and a partner voucher flow. ANODA designed the app and designed and built the website in Webflow, including the Webflow-side connection for voucher redemption; the client’s agency built the app and its backend. The website explains the same music-for-running proposition, then guides partner members through code redemption and toward the app. It connects the public proposition, partner entitlement and first app experience into one launch journey.

For this kind of launch, review the public promise, entitlement explanation, invalid-code path and destination after redemption as one journey. The wider mobile app design task is making sure the first useful experience continues what the website promised.

Start marketing before launch

The worst time to start marketing an app is the day it goes live. You have no audience, no reviews, no rankings and no idea which message works. Pre-launch is where you buy that knowledge cheaply, while changing your mind costs a sentence, not a sprint.

Pre-launch is a real sequence. The order matters because each step feeds the next.

1. Research the audience and the promise. Interviews, reviews of competing apps, forums and support threads where your future users complain. Out of this comes the positioning and the words people use.
2. Put up a landing page with a waitlist. One page, one promise, one call to action. The sign-up rate tells you whether the promise lands before you have built everything behind it.
3. Run a beta. Real users, real devices, real feedback, a small enough group that you can talk to them.
4. Prepare the store pages. Screenshots, preview video, keywords and localized text, tested on people who haven’t seen the app before.
5. Line up launch channels. Your waitlist, your community, press and creators you’ve already talked to, and partners who reach the same audience.
6. Launch, then watch activation, not the install counter.

The landing page and the waitlist

A pre-launch landing page is not a brochure. It is an experiment with one variable: does this promise make this audience raise a hand? Keep it to the promise, a short explanation of how the app delivers it, a visual of the real product, and a single sign-up field. Every extra section is another place for the test to go wrong.

Drive a small amount of traffic from the channels you plan to use later. If a community post brings sign-ups and a paid ad doesn’t, you’ve learned something about your channel mix for the price of a lunch. If nothing brings sign-ups, you’ve learned something far more valuable: the promise needs work, and you found out before building the rest.

Then treat the waitlist as an audience, not a number. Write to it. Ask what made people sign up. Invite the most engaged ones to the beta. A waitlist you only email on launch day is a guest list for a party you never mentioned again. Half of them will have forgotten why they signed up.

Beta programs that teach you something

A beta is the cheapest usability study you will ever run, if you treat it as one. The goal is not buzz. The goal is to watch real people try to reach the value on their own devices and see where they get stuck.

Both stores have built-in routes. On iOS, TestFlight lets you invite up to 10,000 external testers by email or a public link. On Google Play, closed testing lets you distribute pre-release builds; open testing becomes available after production access. If you publish from a personal Play Console account created after 13 November 2023, a closed test with at least 12 testers opted in for 14 continuous days is required before you can apply for production access. Plan for those two weeks. Founders who discover the rule on launch day discover it the expensive way.

Invite-only betas are also a marketing tool. Scarcity makes people pay attention, and a tester who got in early and was listened to is far more likely to recommend the app later. The trick is to keep the group small enough to actually talk to. A beta with 5,000 silent testers teaches you less than one with 40 people you interviewed.

What to watch in a beta:

• Where testers drop out of onboarding, and what they were looking at when they did.
• How long it takes to reach the activation event, and how many never do.
• The words testers use to describe the app to someone else. Those are your future ad headlines.
• What they expected the app to do that it doesn’t. That is either a roadmap item or a promise you need to stop making.

Turn the store page into the best salesperson you have

Almost every install passes through a store page. Paid ads, press, creators, word of mouth: sooner or later, the person lands on your App Store or Google Play listing and decides in a few seconds whether to tap Get. It works like a school open day. Parents walk down the corridor, glance into three classrooms and decide. Nobody reads the prospectus.

App store optimization (ASO) has two jobs, and teams often do only one. The first is visibility: ranking for the searches your audience makes, through the title, subtitle, keywords, description and ratings. The second is conversion: turning a store visit into an install. Visibility without conversion is a shop on the high street with the shutters half down.

What converts on a store page:

• A title and subtitle that state the job. Not just the brand name. The subtitle is where the promise earns its place.
• Screenshots that show outcomes, not features. The first two or three do most of the work, because many people never scroll further. Each one should make one point in a few words, over a real screen.
• A preview video that shows the core action early. Not a logo animation.
• Ratings and recent reviews that answer doubts. Reply to critical reviews. People read the replies.
• Localized text and screenshots. A translated description with untranslated screenshots is half a welcome.

Different pages for different people

One store page for every audience forces a compromise that suits nobody. Both stores now let you split it. On the App Store, custom product pages let you create additional versions of your product page with different screenshots, previews and promotional text, each with its own URL, and Apple supports up to 70 custom product pages, with keywords that can make them visible in App Store search. On Google Play, custom store listings let you tailor the listing by country, by campaign or by other segments, up to 50 per app.

That turns the store page into part of the channel plan. The runner who clicked a running-plan ad lands on screenshots about running plans. The person who searched for strength training lands on screenshots about strength training. Same app, different first impression: match the message to turn the same app into a more convincing offer for each audience.

Test the page, don’t argue about it

Store page debates are a fine way to burn a week. Test instead. App Store product page optimization lets you test up to three alternative versions of your default page’s icon, screenshots and preview videos against the original. Google Play’s store listing experiments do the same job on Android. Test one idea at a time, run the test long enough to mean something, and keep a log of what you learned so the next team member doesn’t re-run your experiments with a new font.

For a deeper dive into rankings and keywords, see our guide to app store ranking and optimization.

Ratings and reviews are marketing you don’t pay for

Ratings sit right next to your install button, and people read them the way they read the hygiene certificate in a restaurant window. A 4.7 doesn’t sell the app. A 3.1 quietly kills every campaign you run, because each visitor you paid to bring in reads the one-star review about the login bug before they read your screenshots.

Reviews are also the most honest research you’ll get for free. Users tell you, in their own words, what they expected, what broke and what they love. That is positioning material, onboarding feedback and a bug list in one place.

What we recommend:

• Ask for a rating at a good moment, right after the user got value, using the stores’ own in-app review prompts. Never after an error, never on first open.
• Reply to critical reviews, briefly and specifically. Future users read the reply more closely than the complaint.
• Fix what reviews repeat. One complaint is an opinion. The same complaint twenty times is a product decision somebody forgot to make.
• Feed review language back into the store page. When users describe the value better than your copywriter did, borrow their words.

A campaign that sends thousands of people to a page with a falling rating is paying full price for a first impression that has already gone wrong.

Localization: new markets without new guesswork

If a meaningful share of your audience reads another language, localization is one of the cheapest acquisition wins you have. A localized store page reaches searches you’re invisible for today, and it converts better because people trust an app that speaks to them properly.

Translate the promise, not just the words. A literal translation of a clever English subtitle often lands as nonsense, and the joke that sells in London can confuse in Madrid. Check the screenshots too: text baked into images is the part everyone forgets, and a translated description over English screenshots is half a welcome.

Then go one step further than the store. If you market in a new language, the onboarding, the key screens, the emails and the push notifications need to follow. Otherwise you’ve built a shop window in the local language and a till that only speaks English. Start with the markets where you already see organic installs or search interest, localize the full first-session journey, and measure activation there separately before you put paid budget behind it.

Choose a small channel mix, not every channel

Here is where most app marketing guides hand you a list of 30 channels and wish you luck. The list is accurate. It is also useless, because no team can run 30 channels well, and a channel run badly teaches you nothing except that it didn’t work when you did it badly.

Spray and pray is not a strategy. It’s a way to spend a budget without learning anything from it.

Loading every channel at once is a school timetable with fourteen subjects crammed into the first week. Nobody learns anything, the teachers are exhausted, and at the end of term you can’t tell which lessons worked because the pupils never stayed in one long enough.

Instead, choose two or three channels and run them properly. Pick them with four questions:

• Where is your audience already paying attention? Not where you’d like them to be. Where they are. Developers on developer communities, parents in parenting groups, finance teams on LinkedIn, gamers on video and streaming platforms.
• What stage is your product at? Pre-launch favours community, content and waitlists. Early growth with a working onboarding can handle paid tests. A mature product with loyal users can lean on referral and cross-promotion.
• What can you afford to run well? Budget, but also people. Content needs a writer. Community needs someone who actually shows up. Paid acquisition needs someone who reads the numbers every week.
• How fast will the channel give you evidence? Paid ads answer in days. Content and search answer in months. You probably need one fast channel to learn from and one slow channel to build on.

A starting mix for a small team often looks like one owned channel (email, content or community), one borrowed channel (creators, partners or press) and, once onboarding is proven, one paid channel for fast learning. That’s it. Add a fourth when one of the three is clearly working and someone has the time to run another one properly.

Channels by lifecycle stage

Channels are not interchangeable. Each one reaches people at a different stage and in a different state of mind, which is why each one needs a different message and a different destination.

Here is how the main channels behave, and what each one needs from the product to work.

Content and search. Articles, guides and landing pages that answer the questions your audience already types into a search engine. Slow to start, compounding once it works, and it builds trust in a way ads never do. It needs pages that match the search intent and lead to a store page that continues the same promise.

Community. Forums, groups, subreddits, Discord servers, Slack communities, local meetups. The rule is simple and constantly broken: contribute before you promote. A founder who answers questions for a month earns the right to mention their app. One who drops a link on day one earns a ban. Community is slow, cheap in money, expensive in time, and the best source of honest feedback you’ll get.

Email. Your waitlist, your beta testers, your existing users. The most reliable owned channel you have, because no algorithm decides whether people see it. It works best when each message has one job and matches where the person is in their journey.

Press. Useful at launch and at real milestones, if you have a story rather than an announcement. “Startup launches app” is not a story. A problem, a surprising number from your own data, a clear point of view: those are. Prepare a simple press kit with screenshots, a short description and a contact, and write to journalists who cover your space, not to everyone.

Creators and influencers. Borrowing someone else’s audience and their trust. It works when the creator’s audience matches yours and the creator actually uses the app. Rented enthusiasm shows. Give creators a clear, honest brief, their own tracked link, and a landing experience that matches what they said on camera.

Partnerships. Other products, communities or organisations that serve the same audience without competing with you. An integration, a bundle, a co-hosted event, a mention in their onboarding. Slow to set up, often cheap, and the audience arrives pre-qualified.

Paid acquisition. App advertising through the store ad networks, search, social and video platforms. Fast, measurable, and the fastest way to learn which messages work, which is why it deserves its own section below.

Referral. Existing users bringing new ones. Powerful and nearly free, but only after users reach the value. Nobody recommends an app they haven’t figured out.

Cross-promotion. If you already have other apps, or a website with an audience, promote the new app to people who already trust you. Match the audiences, though: promoting a finance app to the players of your puzzle game is cross-promotion in name only.

Notice what all of these have in common. Every one of them sends people somewhere, and every one of them fails if the somewhere doesn’t keep the promise.

Match the message, the channel and the destination

A person who clicks a video ad while scrolling on the sofa is in a different state from one who searched “shift swap app for restaurants” at 11pm after a scheduling disaster. The first is curious and has two seconds of patience. The second has a problem and wants proof it’s solved. Send both of them to the same generic home screen and you’ve wasted one of them, probably both.

Sending every campaign to the same default page is putting the cart before the horse: you’ve spent the effort to find exactly the right people, then greeted them all with the same welcome mat. Each channel needs three things aligned:

• The message: what this audience, in this context, needs to hear to take the next step.
• The destination: the store page, custom product page or in-app screen that continues exactly that message.
• The first experience: what they see after installing, which should pick up the story where the ad left off.

Five channels, five contexts, five destinations. The expensive part of marketing is finding the right person at the right moment. Greeting all of them with the same default page throws that work away at the door, and no bid strategy will win it back.

Deep links: land people where you promised

A deep link sends a person to a specific place inside your app instead of its front door. A deferred deep link does the same for people who don’t have the app yet: they go through the store, install, and on first open land on the screen the campaign promised, not on a generic welcome carousel.

Without deep links, campaigns lose people at the worst possible moment. It’s a courier with the right street address and no flat number. The parcel arrives at the building. It sits in the lobby. Nobody knows whose it is.

Picture the user journey. The ad says “Swap your Saturday shift in two taps”. The user taps, installs, opens the app and sees a welcome carousel, a sign-up form and then an empty home screen. The Saturday shift is nowhere. They were promised a two-tap swap and got a four-screen admin task. You paid for that click.

The technical foundation is Universal Links on iOS and Android App Links on Android, which let a normal web link open the right screen in your app when it’s installed. Deferred deep linking, which carries the destination through the install, usually comes from a mobile measurement or deep-linking provider. One warning worth repeating: Firebase Dynamic Links shut down on 25 August 2025, with served links and related APIs retired. If your campaigns, emails or referral invites still use them, they are sending people to an error page right now.

Deep links are not only for ads. Use them in emails, push notifications, referral invites, creator links, QR codes on printed materials and support replies. Every time you send someone into the app, send them to the screen that matches why you’re sending them. And test every link on a clean device, with and without the app installed, before the campaign goes live. That test takes minutes. Skipping it costs the whole campaign.

Onboarding is where most campaigns die

You can do everything above well — positioning, store page, channels, deep links — and still lose most of the people you paid for in the first two minutes. Onboarding is where the marketing promise meets the product, and too often the product answers with a form.

The usual crimes are familiar. A carousel of four slides explaining what the app does, which the user skipped because they already know, that’s why they installed. A sign-up wall before any value. Permissions requested before the user knows why they’d want notifications or location. A profile form asking for things the app doesn’t need yet. Then an empty home screen with a friendly illustration and no clue what to do next.

Each of these is small. Together they are a toll booth on every road you paid to build. A user who came from a campaign promising “swap your shift in two taps” and hits six screens of admin doesn’t blame the onboarding. They blame the ad for lying, and they’re right.

Good onboarding for a marketed app follows the promise:

• Continue the story the channel started. If the user came from a deep link or a specific campaign, start there.
• Get to the activation event as fast as the product allows. Ask only for what is needed for that first value.
• Ask for permissions in context, at the moment the user sees why they help, not on the first screen.
• Delay the account wall where the product allows it. Let people try the core action, then ask them to save their work.
• Design the empty states. The first screen after onboarding should show what to do next, not an empty list.

When installs and sign-ups exist but meaningful use doesn’t, that is not a marketing problem anymore, and more spending won’t solve it. It needs someone to walk the journey from the store page to the core action and find the friction that kills it. That is exactly what our UX audit of activation and retention is for.

Paid acquisition: when to turn it on and how to read it

App advertising is the most tempting lever because it responds immediately. Put money in, installs come out. That’s also what makes it dangerous. It will happily scale whatever funnel you have, including one that loses most people in onboarding.

The main options, at the time of writing:

• Store ads. Apple Ads shows your app in App Store search results and other App Store placements. Google’s app campaigns run across Search, Play, YouTube and its display network.
• Paid social and video. Meta, TikTok, Snapchat, Pinterest and similar platforms, where creative does most of the work and targeting relies increasingly on the platform’s own optimisation.
• Ad networks and programmatic. Broad reach, cheaper inventory, and more work to keep quality under control.

Turn paid on when the readiness gates are passed, not before. Then treat the first budget as tuition, not growth. A few rules we give every team:

• Judge channels on cost per activated user, and later on retention and revenue by cohort. Cost per install on its own will lie to you with a straight face.
• Test messages, not just audiences. The fastest thing paid ads can teach you is which promise makes people act. Feed that back into the store page and onboarding.
• Give each campaign its own destination. Custom product pages and deep links make paid far more efficient than sending everything to the default listing.
• Watch creative fatigue. Ads wear out. Plan a steady supply of new creative rather than one big shoot.
• Don’t scale a channel whose users don’t activate. More budget into a leaky funnel is good money after bad, however attractive the cost per install looks.

One more rule, because we see it break budgets every year: decide in advance what result will make you stop. If you don’t, the test never ends, it just becomes the marketing plan by accident.

Referral and word of mouth

Referral is the channel every founder wants and few earn. The logic is simple: happy users bring similar people, those people trust the recommendation, and the cost is a fraction of paid acquisition. The catch is in the first two words. Happy users. Referral amplifies satisfaction. It can’t create it.

Think of a school party. Kids bring their friends to the second one only if the first one was good. Handing out “bring a friend” vouchers at the door of a boring party doesn’t produce more guests. It produces more people who had a boring time.

What makes referral work:

• Ask at the value moment, right after the user got something good out of the app, not on first open.
• Make sharing useful for the sharer, not just for you. Many apps are naturally better with friends, colleagues or family in them. Lean into that before bolting on rewards.
• Make the invite land properly. A referral link should deep link the friend to a screen that makes sense for them, with context about who invited them.
• Keep rewards honest and simple. If you add incentives, make sure they reward real use, not just installs, or you’ll pay for accounts created to claim the reward.

Word of mouth also happens without any referral programme at all, in reviews, conversations and screenshots. You can’t control it, but you can make it easier: a product that’s easy to explain in one sentence gets explained more often.

Retention: marketing that happens after the install

Retention is where most of the money in app marketing is made or lost, and it’s the stage most marketing plans forget. A user who stays for months is worth many times more than one who leaves in a week, and every one who leaves has to be replaced with another paid install. Every quarter you don’t fix retention, you choose to keep paying for the same user twice.

The pattern is a gym in January. Everyone signs up. The car park is full for two weeks. By February it’s the regulars and a few guilty stragglers. The gyms that do well aren’t the ones with the best January ads. They’re the ones that give new members a reason to come back in week three.

Retention marketing is mostly lifecycle messaging tied to what the user has actually done: email, push notifications and in-app messages that react to their state. Done well, each message helps the user get more value. Done badly, it’s a nagging timer that trains people to turn notifications off.

A simple lifecycle plan covers a few moments:

• Day 0: the user installed but didn’t reach activation. Help them finish, pointing straight to the step they’re stuck on.
• Day 1 to 3: they activated. Show the next most useful thing, the one that turns a single use into a habit.
• Day 7: they’ve gone quiet. Remind them of the value they got, with a deep link back to it, not a generic “we miss you”.
• Day 30 and beyond: they’re regulars. Introduce deeper features, ask for a review, invite them to refer.

Push notifications deserve special care, because they are easy to send and easy to ruin. Ask for permission in context, send fewer and better messages, make each one useful on its own, and let people choose what they receive. We cover the mechanics in our piece on how push notifications work. The short version: a notification that doesn’t help the user is an advert for the “turn off notifications” switch.

And remember that no message fixes a product people don’t want to return to. If users don’t come back after the first week, look at the product before the email copy. Retention messaging reminds people of value. It can’t invent it.

The metrics that belong to each stage

Every stage of the funnel has its own numbers, and each number should support a decision. A metric nobody acts on is decoration. A dashboard full of decoration is a very expensive screensaver.

Two principles make these numbers useful rather than confusing.

Look at cohorts, not averages. Group users by when and where they arrived, and follow each group over time. A channel that brings users who are still active a month later is worth far more than one with a lower cost per install and a cohort that vanishes in a week. Averages mix the two and hide both.

Connect acquisition to activation. The single most useful number for a growing app is often cost per activated user by channel. It’s the one number that makes marketing and product look at the same thing.

Attribution in a privacy-first world

Measuring which channel brought which user has become harder, and it isn’t getting easier. On iOS, App Tracking Transparency means apps must ask permission before tracking users across other companies’ apps and websites, and many people say no. Apple’s privacy-preserving attribution frameworks report campaign results in aggregated, delayed form rather than per user. Check each platform and attribution provider’s current capabilities before deciding what the campaign can measure.

That doesn’t make measurement pointless. It makes it less precise and more dependent on what you own. In practice:

• Instrument your own funnel well. Measure first open, activation and returning use with an event plan that follows your privacy commitments, consent requirements and data-minimisation rules.
• Use store analytics. App Store Connect and the Play Console show store traffic, conversion and, for custom pages and listings, which version performed.
• Treat ad platform numbers as one opinion. Compare them with your own data and look for agreement in direction, not exact matches.
• Run clean tests. Turning a channel on or off in one region, or for one period, and watching activation and revenue can tell you more than any attribution report.

If a mobile measurement partner is part of your setup, choose it for how well it fits your product and privacy stance, not for the prettiest dashboard. The dashboard is not the decision.

When to scale, change or stop a channel

Every channel you run should have a decision date and a rule. Without them, channels don’t get evaluated. They get inherited. Kicking the can down the road is the most expensive marketing strategy there is, because it never shows up as a line item.

Set the rules before you start, while you’re still objective:

• Scale when a channel brings activated users at a cost your model can carry, and cohorts from that channel retain at least as well as your average. Scale in steps and watch whether quality holds as volume grows. It often doesn’t.
• Change when a channel brings the right people but they don’t activate or stay. The audience is right; the message, the destination or the onboarding isn’t. Fix the part after the click before judging the channel.
• Stop when a channel has had a fair test, with a matched destination and a working onboarding, and still brings people who don’t activate. Stop cleanly, write down what you learned, and move the budget.

The hardest one is stop. Teams keep channels alive because they’ve invested in them, because someone on the team likes them, or because “it’s starting to work”. Sometimes the juice isn’t worth the squeeze, and the honest thing is to say so. A stopped channel with a written lesson is progress. A zombie channel with a monthly invoice is not.

A lean plan for teams that can’t do everything

Most teams reading this don’t have a growth department. They have a founder, a designer, a developer or two, and a budget that makes every decision feel personal. Here is a sequence that fits that reality. It’s a real order, because each phase depends on the one before.

1. Before launch: agree the audience and the promise from evidence, define activation, put up a landing page with a waitlist, and run a small beta as a study.
2. At launch: make sure the store page matches the promise, deep links work, onboarding reaches value in the first session, and analytics track the funnel from first open to activation.
3. First month: run two or three channels that fit your audience, including one fast one to learn from. Judge them on activation, not installs.
4. Second and third months: fix the biggest leak inside the product, add custom store pages for the audiences that work, start lifecycle messaging.
5. After that: scale the channel that brings activated, retained users. Add referral once users are clearly getting value. Stop what doesn’t work, and write down why.

It won’t look impressive on a slide. There’s no “omnichannel”, no viral loop in month one, no influencer army. It is the plan that stops you paying for the same user twice, and that tends to be what separates the apps still running in two years from the ones with a great launch week.

A worked example: marketing a shift-swap app

To tie the pieces together, here is an illustrative example. The app, its numbers and its decisions are invented to show the method, not taken from a client.

A small team is building an app that lets restaurant staff swap shifts without a messy group chat. Their first plan was the classic one: launch in both stores, run social ads to “anyone who works in hospitality”, send everyone to the default store page, and count installs.

Before spending, they run the diagnosis. Their positioning is vague (“the scheduling app for teams”), activation isn’t defined, the store screenshots show the settings screen, onboarding asks for a restaurant code, a role and a phone number before showing anything, and campaign links open a welcome carousel.

So they change the order:

• Promise: “Swap a shift in two taps, without the group chat”, for restaurant staff first, managers second.
• Activation: the first swap requested or accepted.
• Pre-launch: a waitlist page shared in hospitality communities and by two restaurant owners they know, then a beta with a few dozen staff from those restaurants.
• Beta lesson: staff drop out at the restaurant-code step. Managers set up the restaurant, so staff shouldn’t have to. The team moves setup to managers and lets staff join from an invite link.
• Store page: first screenshot shows a swap being accepted; a custom page for managers shows the schedule view.
• Channels: community posts and restaurant partners as the slow, trusted channels; a small paid test with two messages, one about swaps and one about fewer group-chat messages.
• Measurement: cost per activated user by channel, day-7 retention by cohort.

In this example, the paid test shows the “fewer group-chat messages” ad brings cheaper installs, but the “swap in two taps” ad brings users who actually swap a shift. The team scales the second one, rewrites the first screenshot around it, and stops the other. None of that is visible if they count installs. All of it is obvious once they count activation.

The app marketing mistakes that waste the most money

Most expensive mistakes in app marketing aren’t exotic. They’re the same half-dozen, made with confidence.

• Buying installs before onboarding works. You pay for people to discover your product is confusing. They don’t come back to see if you fixed it.
• Counting installs as success. The metric goes up, the business doesn’t, and nobody notices until the money runs low.
• Sending every campaign to the same place. Different people, different contexts, one generic first screen. Most of the targeting work is thrown away at the door.
• Running every channel at once. Nothing gets done properly, nothing produces clear evidence, and the team concludes that “marketing doesn’t work for us”.
• Promising what the first session can’t deliver. Ads for features that are hard to find, or not built yet. Every broken promise turns into a review.
• Launching without a beta. The first real users find the problems in public, with ratings.
• Ignoring the store page. Great ads, a store page that looks like a draft, and a conversion rate that quietly doubles your cost per install.
• No stop rules. Channels that should have been cut run for months because nobody decided what failure looks like.
• Forgetting retention. A plan full of acquisition and empty after day one. Everyone you acquire has to be replaced, forever.

The common thread is the one this guide started with. Each of these mistakes treats marketing and the product as separate jobs. The people you pay to bring in don’t experience them separately. They see one app, from the ad to the fifth time they open it, and they judge all of it.

When a promotion plan exists but the journey from store page to core action can’t support its promise, the fix isn’t a better campaign. It’s aligning positioning, the store page, onboarding and the primary mobile journey so the experience keeps the promise the marketing makes. That is the work behind our mobile app design projects: we don’t run your ads, we make sure the people your ads bring in find what they were promised.