App Installs in Affiliate Marketing: CPI, Attribution, and Traffic Quality

What App Installs Mean in Mobile Advertising and Affiliate Marketing: Install Types and CPI

App installs are application installations, usually from official or unofficial app stores, also called marketplaces or app marketplaces. The most common are Google Play and the Apple App Store. There were or still are Amazon Appstore, Windows Phone Store, and BlackBerry World, but by now they are gone or mostly exotic, although you can still find an affiliate angle somewhere. In affiliate marketing and mobile traffic, this is an offer type where the advertiser wants installs and pays either for an install or for a conversion inside the app.

Post updated August 11, 2026.

How Relevant Install Offers Are for Affiliates in 2026

I would split this into three parts:

• Installs for iGaming gambling and betting offers are less relevant, but you can still find them. This is not the best time for installs in these categories: PWA solves most of the pain perfectly well.
• Non-gambling installs through affiliate networks (in Russian). The old market where you could just pick a mainstream app and start working with it easily is gone. There is assorted crypto stuff (simulators), utilities, and VPNs with absurd pricing. Horoscopes and NSFW apps appear now and then, but they are exceptions.
• Your own offers became relevant in early 2026 as AI developed. You can vibe-code or vibe-engineer an app, or build a relatively fast app-production process for products slightly more complex than WebView wrappers. On one hand, app marketplaces are the reddest of red oceans. On the other, if you do not annoy the marketplace, it gives you excellent user billing options. Affiliates are used to red oceans and can find ways to acquire installs for their own apps. The question is the monetization balance: aggressive enough to make money, but with enough quality not to get kicked out of the store.

The most common marketplaces for affiliates:

• Google Play: app installs for Android
• Apple App Store: obviously, installs for iOS devices

The flow is very simple for the user: they click a banner, or less often a link from a prelander, and land in one marketplace or another, where they install the app through the marketplace interface. Obviously, users trust the marketplace quite a lot.

What Makes App-Install Campaigns Different?

Several parameters have a major effect on CR:

• Download size. The heavier the app, the less willing users are to install it and the lower CR can be. There has not been a universal 30 MB threshold for a long time. Google Play shows a warning when an app over 200 MB is downloaded over a mobile network, but does not block the installation. Measure the real effect of size by store, GEO, device, and CR instead of guessing from an ancient chart.
• Description and screenshots. Sometimes the description is frankly bad. Sometimes not very competent ASO people work on it, and afterward nobody wants to read the description or install the app. The same applies to screenshots. In those cases, try a prelander. Installs are usually promoted directly to the stores, but in my experience a prelander produces more profit about half the time.
• How new or exhausted the install offer is. This is simple: the audience may already have been hammered, and everyone who wanted the app has installed it. The newer the install offer, the more likely it is to take off.

What Counts as an Install in 2026

A file on the phone is not always a paid install. In a CPI funnel, the network or MMP usually waits for first open and checks whether it can be attributed to a click or impression within the attribution window. A reinstall, redownload, duplicate, or user already in the database may not count. The exact rules are in the offer terms, not floating in the air.

On iOS, part of attribution uses privacy-preserving postbacks, including AdAttributionKit and SKAdNetwork. On Android, the available signals depend on Google Play, the ad platform, and the MMP. The conclusion for a media buyer is simple: before launch, ask what counts as an install, when it appears in statistics, how reinstalls and duplicates are filtered, what the attribution window is, and which post-install events are required. Otherwise, you can buy real installs and get paid exactly zero.

Metrics and Factors When Working with App Installs

• Retention Rate is the share of users who continue using the app. For example, RR7 of 30% means that 30% of the Day 0 cohort remains on Day 7. For apps built around long LTV, this is the main metric. For aggressive monetization and one-time billing, it matters less.
• Lead cap is a limit on installs for the entire network or for a particular publisher. Caps are usually low, in the hundreds per day. Even if they said unlimited at first, after 1,000 installs they may ask you to stop traffic. Do not drag out optimization: you can reach break-even just when every cap is filled.
• Conversion Rate. Obviously, almost everyone now looks at in-app events and calculates LTV, RPU, and other useful metrics. Even so, CR remains the simplest and fastest signal for spotting incent, bots, other trash, and fraud. The people responsible for network relations get instructions from above: “If CR > n%, stop traffic and say it is incentivized.” This is bad, but those are the rules of the game for now. What do you do? The usual: talk to the manager and ask what they want. They may allow CR above the system threshold or ask you to dilute it.
• App lifecycle stage. If it is a soft launch, you probably will not see the offer at all. If you do, a soft launch always needs high-quality, stable, controlled traffic. If it is a boost, it will probably need incentivized traffic, especially for iOS installs. In that case, the advertiser is ready to accept traffic of almost any quality, as long as it gets the required number of installs in time to reach the top of the store. The clearest and most interesting stage for an affiliate is building a real user base, after the advertiser has already exhausted the obvious high-quality sources and started placing offers in networks. The advertiser still expects quality traffic here, but the requirements are somewhat lower than at soft launch. Such offers can live for several months.
• Advertisers now have another pile of important metrics. The more mainstream the app and the more it depends on long-term LTV, the more these metrics matter: activation, D1/D7/D30 retention, target in-app events, N-day ROAS, LTV, reinstalls and duplicates, fraudulent installs (in Russian). Together, these metrics define install quality and the rejection rate. See other definitions in the CPA Glossary.
• For apps with referral mechanics, add the K-factor: how many new users each current user brings.

That old recommendation is outdated in 2026: you can start with installs, but you need a fairly good grasp of how players behave and you need to grill managers hard about what is worth running right now. I would say that creating your own install offers is a good way to diversify if you have the skills and resources, but it is definitely not a beginner story.

I am not covering incentivized installs or .apk installs here. They are interesting topics, but they need separate posts.