How to Measure App Conversions with Multi-Touch Attribution
How app marketers use multi-touch attribution, deep links and web-to-app journeys to understand installs and conversions, and what Apple's privacy rules changed.
For an app, a conversion is rarely one step. Someone sees a social ad, visits the mobile site, gets a reminder email, installs the app a week later and makes a first purchase a week after that. Crediting only the last tap before the install misses most of what happened.
Multi-touch attribution (MTA) is how app marketers try to see the whole path. It has become harder to do well since Apple's privacy changes, which makes it more important to know what it can tell you and what it can't.
Why last-touch attribution isn't enough for apps
Last-touch attribution gives 100 percent of the credit for an install to the final interaction, often an app store search or a retargeting ad. Channels that introduced the app, or brought people back to it, get nothing.
Multi-touch attribution spreads credit across the touchpoints along the way. For apps, it also follows users past the install to the events that matter more: registrations, subscriptions, purchases and retention. A channel that drives cheap installs from people who never return looks very different once those later events are counted.
How app attribution is measured
Most app marketers rely on a mobile measurement partner (MMP), a platform that collects ad interactions from many networks and matches them to installs and in-app events. These platforms typically bundle several tools:
- Attribution across ad networks, owned channels and the web.
- Deep linking, which sends people to the right screen in the app rather than the home screen.
- Analytics on installs, in-app events and retention by campaign.
- Audience building for re-engagement campaigns.
- Fraud protection against fake installs and click spam.
Some also add incrementality models, which compare people who were exposed to a campaign with similar people who weren't, using statistical matching methods such as propensity score matching, to estimate the extra conversions a channel produced. That's a different question from attribution, and a more useful one for budgeting.
What Apple's privacy changes did to app attribution
App attribution used to depend on device identifiers, like Apple's IDFA, that could be matched across apps and networks. Apple's App Tracking Transparency (ATT), introduced in 2021, requires apps to ask users for permission before tracking them across other companies' apps and websites. Most people say no.
For the users who decline, device-level, cross-app attribution is no longer possible. Apple offers its own frameworks instead, SKAdNetwork and its successor AdAttributionKit, which report installs and limited post-install signals in aggregated, delayed and privacy-protected form. They confirm that a campaign produced installs, but with far less detail and no user-level paths.
Android still offers an advertising ID, though users can reset or delete it, and cross-platform journeys remain hard to stitch together. The practical result: app MTA now sees a fuller picture on some users and platforms than others, and its reports should be read as a partial view.
Web-to-app journeys
As cross-app tracking has shrunk, the journey from your own website into your app has become more valuable, both for conversion and for measurement.
- Smart app banners prompt mobile web visitors to open or install the app.
- Deep links take people to the exact product, offer or screen they were looking at, even if they need to install the app first.
- Consistent handoffs carry campaign information from the web visit into the app, so the conversion stays connected to the campaign that started it.
These journeys reduce drop-off between web and app, and because they run through your own properties, they keep more of the path measurable without relying on cross-app tracking.
Getting more from app attribution
A few practices make app attribution more reliable and more useful:
- Optimize to downstream events, not installs. Judge channels by registrations, subscriptions, revenue and retention, not by cost per install.
- Compare models. If a channel's credit changes sharply between last-touch and multi-touch views, it's probably assisting rather than closing.
- Know your coverage. Check what share of conversions can be attributed at the user level and what share comes only through aggregated reporting. Decisions based on the visible share can be skewed.
- Watch retargeting and brand search. These channels reach people who already know the app, so they tend to collect attribution credit for conversions that would have happened anyway.
- Allow for reporting delays. Apple's aggregated install reports arrive days after the install, so a campaign judged in its first few days will look weaker than it is. Wait for the reporting windows to close before cutting it.
- Use the same event definitions everywhere. A "purchase" or "trial start" should mean the same thing in your MMP, your ad platforms and your own analytics, or the numbers will never reconcile.
Pair attribution with incrementality
Attribution answers "what did users touch before converting?" Budget decisions need the answer to "what would we lose if we stopped this channel?"
That's the job of incrementality measurement. Marketing mix modeling works from aggregate spend and outcome data, so it doesn't depend on device IDs or user consent, and it can include the channels that drive app demand without a tap, such as connected TV, audio and out-of-home. Many app marketers now use MTA for day-to-day campaign optimization and MMM or controlled tests to set budgets across channels.
Used that way, multi-touch attribution remains a valuable tool for app growth. It just shouldn't be the only one.
Frequently asked questions
How does multi-touch attribution work for mobile apps?
It connects the ads, web visits and messages a person interacted with to the install and to later in-app events like purchases or subscriptions. Mobile measurement partners collect this data across ad networks and assign credit across the touchpoints instead of giving it all to the last one.
How did App Tracking Transparency change app attribution?
Since 2021, iOS apps must ask permission before tracking users across other companies' apps and sites. Most users decline, so the device-level data that app attribution depended on is missing for much of the iOS audience. Apple's own frameworks, SKAdNetwork and its successor AdAttributionKit, report installs in aggregated, delayed and limited form instead.
What is a web-to-app journey?
It is the path from a mobile website or web ad into the app, using smart app banners and deep links that open the app on the right screen, or send people to the app store and then to that screen after install. A good journey raises conversion and keeps the link between the web visit and the app event intact.
What is the difference between attribution and incrementality for apps?
Attribution credits the touchpoints that came before an install. Incrementality estimates how many installs or purchases would not have happened without a channel. Retargeting and brand search ads often receive attribution credit for users who would have converted anyway.