What is performance marketing for apps?
Performance marketing for apps is any paid marketing where you can count what the money brought in, and where the price is tied to that count. The result can be an install, a trial, a purchase, a paid subscription or revenue. The opposite is brand marketing, where you pay for reach and judge it later by surveys or overall growth.
Three numbers run most of it. Cost per install (CPI) is spend divided by installs. Cost per action (CPA) is spend divided by an in-app action, such as a trial or a purchase. Return on ad spend (ROAS) is revenue from those users divided by spend. Example: $1,000 brings 400 installs and 20 paying subscribers, so CPI is $2.50 and the cost per paying user is $50.
Mobile apps add two problems that websites mostly don't have. The install happens in an app store you don't control, and privacy rules on iOS limit what any ad platform can report about one person. Both shape how every channel below is priced and measured.
What do you pay for in app performance marketing?
You can pay per impression, per tap, per install, per in-app action, toward a revenue target, or as a share of revenue. The further down that list, the closer the price is to money in your account, and the harder it is to measure who caused it. Most large ad platforms let you choose the goal and then bid toward it.
| Model | You pay for | Where you see it | Who carries the risk |
|---|---|---|---|
| CPM | Each 1,000 impressions or views | Impression-priced ad buys; per-view creator campaigns | You: an impression may lead to nothing |
| CPC or CPT | Each click or tap | Apple Ads Advanced charges per tap (Apple) | Shared: a tap is intent, not an install |
| CPI | Each install | Apple Ads Basic charges per install; Google App campaigns offer target CPI (Apple, Google) | Mostly the seller, but installs can be faked or never opened |
| CPA or CPE | An install plus a chosen in-app action or event | Google target CPA (Google); Meta optimizes for app events (Meta) | Mostly the seller; you still pay the platform's bid |
| ROAS-based bidding | Spend steered toward a revenue target | Google target ROAS, Meta minimum ROAS, TikTok target ROAS (Google, Meta, TikTok) | Shared: needs revenue events sent back from the app |
| Offerwall or incentive | A user finishing a task for an in-app reward | Offerwalls inside other apps (AppsFlyer) | The seller, but rewarded users may not stay |
| Revenue share or affiliate | A cut of what the users you got pay | Affiliate and some creator deals | The seller, who waits to be paid |
Read the CPI and CPA rows carefully. A target CPI or CPA tells the platform what to aim for; it is not always the unit you are billed in, so check each platform's billing terms. Apple Ads Basic is one setup that bills per install directly (Apple). For a deeper look at action-based pricing, see CPA app marketing.
Which channels do app performance marketers use?
The main channels are app store ads, the big self-serve ad platforms (Meta, TikTok, Google), ad networks and DSPs that buy space inside other apps, affiliates and offerwalls, and creators. Most apps start with one or two and add more once they can measure cost per result in each.
| Channel | How you usually pay | Strength | Watch out for |
|---|---|---|---|
| Apple Ads (App Store) | Per tap (Advanced) or per install (Basic, up to $10,000 per app a month) (Apple) | People are already in the store | iOS only |
| Google App campaigns | Bids toward installs, in-app actions or value; target CPI, CPA or ROAS (Google) | Runs across Search, Google Play, YouTube, Display and Discover (Google) | Google, not you, spreads ads across those surfaces |
| Meta Advantage+ app campaigns | Optimizes for installs, app events or purchase value; bid cap or minimum ROAS (Meta) | One campaign optimizes toward the goal you pick | Event and value goals need app events sent back |
| TikTok app campaigns | App promotion objective; value-based optimization with Highest Value or Target ROAS (TikTok) | Short-video ads that can optimize for purchase value | Value optimization needs revenue data sent back |
| Ad networks and DSPs | Varies by network and deal | Ad space inside other apps | DSP fraud rate rose from 5.6% to 8.9% in a year (AppsFlyer) |
| Affiliates and offerwalls | Per install, per task, or a revenue share | You pay only when the task is done | Affiliates ran about 40% fraud, against about 1% for self-reporting networks (AppsFlyer) |
| Creators (TikTok, Instagram Reels) | Flat fee, per view, per install or action, or a mix | Real people showing the app; many hooks tested at once | Many viewers search the store instead of tapping a link |
Self-reporting networks are the big platforms that report their own results to your measurement tool, such as large social and search platforms. For what each channel costs in total, including agencies and tools, see how much it costs to market an app.
How do creators fit as a performance channel?
Creators become a performance channel when you pay them for something you can count, not a flat fee per post. That count can be views, installs, trials, paid subscriptions or revenue. Performance-based creator marketing, also called results-based or outcome-based influencer marketing, mixes these so each side carries part of the risk.
- Per view. A rate per 1,000 counted views. Easy to verify, and the creator is paid for real attention. See pay-per-view influencer marketing.
- Per install. A price for each install tied to the creator. Closest to results, but searchers are hard to credit. See pay-per-install influencer marketing.
- Per action. A price per trial, signup or purchase. See CPA app marketing.
- Performance UGC. Creator videos made to run as ads or to be judged on results, not just posted. See performance UGC.
- Hybrids. A view rate for reach, plus a bonus per install, trial or paid subscription you can track.
The catch is measurement. Airbridge, an attribution company, says the main behavior on social apps is watch, then search, and that most users never enter a promo code. AppsFlyer says influencer campaigns typically lift "organic" users 30-40% above the baseline, because those viewers arrive through store search. A pure per-install deal pays creators only for the part a link can see, so many creators turn it down.
For the full picture on models, rates and briefs, see influencer marketing for apps. To turn any flat fee into a price per 1,000 views, use the influencer CPM calculator.
How is app performance measured on iOS and Android?
An app that buys from several channels can use a mobile measurement partner (MMP), a tool that matches ad interactions to installs and later in-app events so one neutral record decides which source gets credit (AppsFlyer). Under it sit the platforms' own tools: Apple's privacy-safe attribution on iOS, Google Play's install referrer on Android, and the stores' campaign links.
| Tool | Platform | What it tells you | Limits |
|---|---|---|---|
| Mobile measurement partner (MMP) | iOS and Android | Which channel gets credit for each install and in-app event (AppsFlyer) | Still bound by Apple's privacy rules on iOS |
| SKAdNetwork and AdAttributionKit | iOS | Signed postbacks to the ad network, with no user- or device-specific data (Apple, Apple) | Values appear only above Apple's privacy threshold; detail is limited for crowd anonymity |
| Google Play Install Referrer API | Android | The referrer URL, click and install times for each install (Android Developers) | Needs code in the app to read it |
| App Store Connect campaign links | iOS | First-time downloads within 24 hours of a tap, plus usage, sales and subscriptions per campaign (Apple) | A metric shows only once it reaches 5; searchers are not counted |
| Google Play links with UTM tags | Android | Store listing visitors and acquisitions by utm_source and utm_campaign (Play Console Help) | A search for your app's name counts as Google Play search, not as your link |
Apple now points app ad campaigns to AdAttributionKit, which works alongside the older SKAdNetwork (Apple). Neither needs App Tracking Transparency permission, but both trade detail for privacy: you learn how a campaign did in aggregate, not what one user did.
What does a paid install cost in 2026?
Worldwide, a paid install cost about $3.03 on iOS and $0.76 on Android in Q1 2026, or $1.24 blended, per Singular. In the US both were higher: $4.60 on iOS and $5.09 on Android. Category matters too: a US finance install on iOS cost $18.06 in the same data.
| What | Platform or market | Cost per install | Source |
|---|---|---|---|
| All apps, global, Q1 2026 | iOS / Android / blended | $3.03 / $0.76 / $1.24 | Singular |
| All apps, United States, Q1 2026 | iOS / Android | $4.60 / $5.09 | Singular |
| Finance apps, global, Q1 2026 | iOS / Android | $14.97 / $2.00 | Singular |
| Finance apps, United States, Q1 2026 | iOS | $18.06 | Singular |
| Mobile games, global, 2025 | Blended | $0.56, up 30% year over year | Adjust, via PPC Land |
| Mobile games by genre and country, Q3 2026 | iOS | From $0.13 (hypercasual, India) to $24.45 (casino, US) | AppsFlyer |
| Mobile games by genre and country, Q3 2026 | Android | From $0.03 (hypercasual, simulation and action, India) to $20.86 (casino, UK) | AppsFlyer |
These are vendor benchmarks from apps that spend heavily on ads, not prices you will get. A cheap install is worth little if nobody opens the app twice, so set your target from what a user is worth to you, not from a table. We found no public benchmark for cost per install from creator campaigns.
Performance marketing for mobile games leans hardest on these numbers: genre and country can move the price of an install by hundreds of times, as the AppsFlyer rows show. Judge game campaigns by revenue per user against cost (ROAS), not by CPI alone.
What are the honest pitfalls of app performance marketing?
Four problems show up again and again: attribution gaps, viewers who search the store, fraud, and optimizing for the wrong number. None of them goes away with a better dashboard; you plan around them.
- Attribution gaps. On iOS, Apple's attribution sends postbacks with no user- or device-specific data, and less detail when numbers are small (Apple). Expect to see results by campaign, not by person.
- Viewers who search the store. Links and codes miss people who watch and then search by name (Airbridge). On Google Play, those installs show up as Google Play search (Play Console Help). Treat link counts as a floor.
- Fraud. In AppsFlyer's 2026 data, affiliates ran about 40% install fraud, against about 1% for self-reporting networks (AppsFlyer). Fake installs can also be passed off as organic: organic traffic made up 52% of fraudulent installs (AppsFlyer).
- The wrong number. A low CPI from users who never open the app again costs more than a higher CPI from users who pay. Judge channels by cost per trial, paying user or ROAS once you have the data.
- Creator fraud. In per-view creator deals the risk moves from fake installs to fake views, so each video needs a view check before money moves.
How do you start performance marketing for a new app?
Start small, measure before you scale, and pick the number you will judge by before you spend. A first test should answer one question: which channel brings a user who does the thing you care about, at a cost you can afford.
- 01
Pick your number
Cost per install early on, then cost per trial or paying user, then ROAS once you have revenue data.
- 02
Set up measurement
Campaign links in App Store Connect and Google Play at the least. Add an MMP once you buy from several paid channels at once.
- 03
Write down your baseline
Note your normal daily installs for two weeks, so you can see lift that no link catches.
- 04
Test one or two channels
For example, Apple Ads for people already searching, and creators for people who don't know the app yet.
- 05
Compare and move money
After the test, put more into the channel with the lowest cost per result you trust, and keep checking it.
Want to compare creator platforms side by side? See the best creator platforms for apps.
Where does Markapping fit in performance marketing for apps?
Markapping is a performance-based creator marketing platform for mobile apps: app owners pay creators for results, not flat fees, at a rate per 1,000 views on TikTok and Instagram Reels plus optional bonuses per install, trial or paid subscription. It is not live yet. App owners and creators can join separate waitlists now, and it runs in the browser with no app to install.
- 01
Set the rate and bonuses
A rate per 1,000 views, from $1.00, which can only go up while the campaign is live. Optionally add a bonus per install, per trial started and per paid subscription, each priced on its own.
- 02
Fund it up front
A budget from $500 plus a 10% platform fee, paid before launch. Unspent budget comes back as a refund or credit, and creators keep 95% of what they earn.
- 03
Creators post with their own link
On their own TikTok or Instagram Reels accounts, with the paid partnership label on. Each creator gets their own store link.
- 04
Results come from the stores
Installs per link are counted by Apple through App Store Connect campaign links and by Google Play. On iPhone, Apple also counts trials and paid subscriptions, with no SDK. On Android, trials and subscriptions need a short code in your app that reads Google Play's install referrer.
- 05
Checks, then payment
Each video counts for 7, 14 or 30 days and gets a fake-view check; a person makes the final call, and the creator can appeal once. Bonuses are paid a few days later, when the store data arrives.
- Viewers who search the store by name are not credited to a link, so link results are a floor, not the full effect.
- You can reject a video only for breaking a written campaign rule. Videos not reviewed within 48 hours are approved automatically.
- Caps limit what any one video or creator can earn.
Markapping does not replace ad platforms or an MMP. It is one channel: creators paid for views, with bonuses for results you can count. See how Markapping works for app owners and pricing.
Questions
What is performance marketing for mobile apps?
It is paid marketing priced and judged by measurable results: installs, in-app actions or revenue. You buy through channels like Apple Ads, Google, Meta and TikTok app campaigns, ad networks, affiliates and creators, and compare them by cost per install, cost per action or return on ad spend.
What is the difference between CPI and CPA?
CPI is the cost of one install. CPA is the cost of an install plus a chosen in-app action, such as a trial or a purchase. Google App campaigns offer both as targets: target CPI for installs and target CPA for an install plus an action (Google).
What is a good cost per install in 2026?
It depends on platform, country and category. Worldwide in Q1 2026, a paid install cost about $3.03 on iOS and $0.76 on Android, and about $4.60 and $5.09 in the US (Singular). A good CPI is one below what an average user pays you over time.
Do I need a mobile measurement partner?
Not at the start. App Store Connect campaign links and Google Play's acquisition reports cover a first test. An MMP helps once you buy from several paid channels, because it gives one neutral record of which source gets credit (AppsFlyer).
Is creator marketing a performance channel?
It can be, when you pay for something you can count: views, installs, trials, paid subscriptions or revenue. Many installs from creator videos come from viewers who search the store later, so link-based counts are a floor. Hybrid deals, a view rate plus result bonuses, share that risk.
Can I pay creators only per install?
You can offer it, but many creators decline, because installs from viewers who search the store can't be tied to their link. Markapping does not work that way: you pay a rate per 1,000 views, and can add optional bonuses per install, trial or paid subscription. See pay-per-install influencer marketing.
How does performance marketing for mobile games differ?
Install prices swing much more. In Q3 2026, AppsFlyer data showed game installs from $0.03 on Android in India to $24.45 for casino games on iOS in the US (AppsFlyer). So judge game campaigns by revenue per user against cost, not by install price alone.
Is Markapping live?
Not yet. Markapping is opening soon, and app owners and creators can join separate waitlists now. It will run in the browser, with no app to install, for iOS and Android apps on TikTok and Instagram Reels.
Sources
- Singular: Quarterly Report 2026 Q2 (Q1 2026 data)
- AppsFlyer: Gaming app market benchmarks for Q3 2026 (Nov 5, 2026)
- PPC Land: Adjust's 2026 mobile app report, gaming CPI up 30% (Feb 22, 2026)
- AppsFlyer: The State of Fraud for Marketers, 2026 Edition (Q1 2025-Q1 2026 data)
- AppsFlyer: Organic traffic is now the single largest ad fraud channel (Jun 10, 2026)
- Apple Ads: Compare Apple Ads solutions (checked Nov 15, 2026)
- Google Ads Help: About bidding in App campaigns (checked Nov 15, 2026)
- Google Ads Help: About App campaigns (checked Nov 15, 2026)
- Meta for Developers: Advantage+ app campaigns (checked Nov 15, 2026)
- TikTok Ads Help: About Value-based Optimization for app (updated Oct 2026)
- Apple Developer: AdAttributionKit documentation
- Apple Developer: SKAdNetwork documentation
- Android Developers: Google Play Install Referrer API
- Apple Developer: App Store Connect campaign links
- Play Console Help: Understand and grow your app's user base
- AppsFlyer glossary: Mobile measurement partner (MMP)
- AppsFlyer glossary: Offerwall
- AppsFlyer: Influencer marketing for apps (updated Jun 2, 2026)
- Airbridge: Influencer attribution, promo code vs tracking link (Mar 24, 2026, updated Nov 5, 2026)