Mobile marketing analytics runs on three layers that keep getting confused: MMP attribution refereeing install credit, the store consoles reporting acquisition and technical health, and product analytics measuring what users do after the install. In this guide we explain all three layers: the store consoles decoded, the in-app layer, the metrics that outrank installs, and the budget layer above all of it.
What are the three layers of mobile measurement?
Layer one: MMP attribution. Which ad network deserves credit for the install, refereed by a mobile measurement partner: AppsFlyer, Adjust, Branch, Kochava or Singular. The full mechanics live in mobile and app attribution, and the field beyond AppsFlyer is compared in our four-MMP guide. Many advertisers face measurement challenges due to their MMP setup.
Layer two: the store consoles. Apple’s and Google’s own reporting on how the store listing performs and how the app behaves technically: the acquisition-and-quality layer, owned by the platforms. Some MMP providers report similar metrics to the platforms but it’s not always the case. The store consoles are very important to frequently check on since they provide many information around the app’s health metrics.
Layer three: product analytics. What users do after installing: activation, retention, revenue. The layer where installs become value and both product and marketing teams try to get business value and optimize.
The layers are not redundant; they answer different questions. Credit between networks is layer one’s job, listing performance is layer two’s, and user value is layer three’s. Teams that conflate them end up asking each tool questions it was never built to answer, and concluding the tools are broken when the taxonomy was.
What are the store consoles for Apps?
There are two major store consoles.
Apple’s App Store Connect includes Analytics: how people discover and interact with your app, covering downloads, conversion rate, retention, monetization metrics, and peer group benchmarks that place your numbers against comparable apps.
Google Play Console’s named features include Statistics, Android vitals for technical quality, crashes and performance as the store experiences them, acquisition reporting on how users reach the listing, and subscription reporting.
Listing views, conversion to install, the technical health signals that quietly shape ranking: this is the ground truth, straight from the source and you should mainly trust your store console as single source of truth. It is blind to everything before the store (which ad drove the visit is layer one’s question) and most of what happens after the install (layer three’s).
The store console is not a measurement platform. Apple and Google will label each activity to the touchpoint they can report successfully. That means that in many cases the measurement problem starts with the console.
The in-app layer: product analytics for apps
After the install, measurement moves inside the app.
Google’s answer is Google Analytics with GA4 properties: event-based, collecting website and app data together so one journey spans both (Firebase’s documentation names the app analytics layer simply Google Analytics). Every action logs as an event, and funnels, cohorts and revenue views build on the stream.
The independent product analytics tools, Amplitude and Mixpanel most prominently, do the same job with deeper product-analytics tooling: event streams, funnel analysis, cohort and retention views. Setting up those in your App requires SDK and your developer should be very familiar with setting it up. Similar to your console, your analytics solution will not improve or fix your measurement problem. If Apple report 80% of app downloads as “Direct” then your Mixpanel will do the same. Mixpanel has the ability to tell you that out of the 80% of users downloaded the app directly 5% purchased within the app.
The MMP side has grown its own analytics: Adjust, for instance, ships Datascape, its analytics product, with all its plan tiers, sitting alongside attribution, deep linking and fraud prevention.
The metrics that matter for apps
Cost per install is a price tag not a verdict. It tells you what a user cost, and nothing about whether they were worth it, and the industry’s long romance with CPI has funded a great many cheap installs that never opened the app twice. Similarly if you advertiser in several channels the cost per install most probably is not accurate.
Activation rate: the share of a cohort reaching first value, the milestone where the app has actually delivered something. An install that never activates is a cost wearing a growth metric’s clothes.
Retention curves, by day and week, by cohort: the shape of coming back. A curve that flattens means a durable base; one that slides to zero means a leaky bucket, however impressive the install count filling it.
Revenue per user, by cohort and by channel: what users actually become worth, which is the only fair basis for comparing the networks that delivered them.
Payback window: how long until a cohort’s revenue covers its acquisition cost, the cash-flow truth under every scaling decision.
The buying rule that follows: judge networks on cohort value, not install volume. You are not buying installs; you are buying the users inside them.
The budget layer above the app funnel
The three layers referee credit, report the store, and measure the users. None of them answers the budget question: were the installs incremental, and how does app spend interact with web, brand and offline marketing? An MMP referees between ad networks; it cannot say whether users would have found the app anyway, and post-ATT the visibility gaps have only widened: cookieless and ATT-era attribution covers that story.
The aggregate layer answers it. In case you are interested in long-term advertising effect then you need a Marketing mix modeling which reads spend and outcomes across the whole business, app and everything around it. In case you care about budget allocation and the interplay of your marketing mix (cross channel) then you need a data-driven attribution solution. MarSci published a peer-review method named: Causal-Driven Attribution. It estimates each channel’s causal influence with no user-level data at all, which makes it naturally at home in mobile’s privacy era: see MarSci’s data-driven attribution solutions.
Frequently Asked Questions
Three layers working together: MMP attribution refereeing install credit between ad networks, the store consoles reporting listing performance and technical quality, and product analytics measuring activation, retention and revenue after the install. Each answers different questions, and the confusion between them causes most mobile measurement arguments.
Apple’s App Store Connect includes Analytics covering discovery, downloads, conversion rate, retention, monetization metrics and peer group benchmarks. Google Play Console offers Statistics, Android vitals for technical quality, acquisition reporting and subscription reporting. Both are authoritative for store behavior and blind to what happens before the store or deep inside the app.
Datascape, Adjust’s analytics product, ships with all its plan tiers alongside attribution, deep linking and fraud prevention. Adjust is an AppLovin company, and Datascape represents the MMP-side analytics pattern: measurement that sits between install credit and full product analytics, reading campaign performance from the attribution data.
Activation rate, the share of a cohort reaching first value; retention curves by day and week, whose shape separates durable bases from leaky buckets; revenue per user by cohort and channel; and payback windows. Cost per install is a price tag, not a verdict: judge networks on what their users become.
MMP reports numbers from your console. Therefore it is difficult for your MMP to know how actually users found your app and download it. Research suggests that upper funnel activity can be the reason.
