Contents
- From gut feeling to steering: what app analytics is for
- The key app KPIs by lifecycle stage
- North star metric: one number that reflects value
- Creating a tracking plan in five steps
- Tools: from Firebase to self-hosting
- Privacy: consent, ATT and data minimisation
- Common app analytics mistakes
- Conclusion
- Frequently Asked Questions
The app analytics KPIs that matter answer three questions: where do your users come from, do they experience the app’s value, and do they stay long enough for the app to pay off? Teams that watch dozens of metrics at once quickly lose focus. This article shows which KPIs really count, how to capture them cleanly and how to stay on top of privacy and consent.
From gut feeling to steering: what app analytics is for
Analytics is not an end in itself. Every metric should be able to drive a decision: is a marketing channel worth it? Is the new onboarding better than the old one? Which feature is barely used and could be simplified? Metrics that never trigger a decision are vanity metrics – they look good but steer nothing. The classic example is total downloads since launch: the number only ever goes up and tells you nothing about whether the app works.
A proven approach is to organise metrics along the user lifecycle: acquisition, activation, retention, revenue and referral. Add technical quality metrics on top, because crashes and loading times affect every other number.
The key app KPIs by lifecycle stage
| Stage | KPI | Definition | What you use it for |
|---|---|---|---|
| Acquisition | Store page conversion rate | Installs ÷ store page views | Impact of screenshots, copy and icon |
| Acquisition | Cost per install (CPI) | Ad spend ÷ attributed installs | Comparing channels |
| Activation | Activation rate | Share of new users who complete the core action | Evaluating onboarding |
| Activation | Time to value | Time from install to first core action | Finding friction at the start |
| Retention | Day-1/7/30 retention | Share of a cohort still active after X days | Comparing retention across releases |
| Retention | DAU/MAU | Daily active ÷ monthly active users | Gauging usage frequency |
| Revenue | Paid conversion | Share of users with a purchase or subscription | Testing paywall and price |
| Revenue | ARPU / LTV | Revenue per user per period, or over the whole lifetime | Deriving a maximum acquisition budget |
| Quality | Crash-free sessions | Share of sessions without a crash | Monitoring release quality |
| Referral | Average rating, invitations | Store rating, invitations sent | Satisfaction and viral growth |
Not every app needs every metric. An internal employee app has no CPI and no revenue, but activation and usage frequency are crucial. A subscription app, by contrast, must know its churn and LTV precisely.
North star metric: one number that reflects value
Many teams work with an overarching north star metric that captures the core value for users and the business at the same time. For a booking app it might be “completed bookings per week”, for a learning app “lessons completed per active user”, for a field service app “jobs recorded digitally”. A good north star metric meets three criteria:
- It only rises when users actually get value.
- The product team can influence it.
- It is a leading indicator of revenue or efficiency, not the outcome itself.
The other KPIs then serve as diagnostics: if the north star stalls, activation, retention or acquisition show where the problem lies. Our article on app user retention strategies explains how to work on retention specifically.
Creating a tracking plan in five steps
The most common reason for unusable analytics is not the tooling but a lack of planning. Events get named ad hoc, captured twice or implemented differently on iOS and Android. A tracking plan prevents this:
- Collect questions: which decisions do you want to make over the coming months? The metrics follow from these.
- Define events: for each metric, specify the events required, e.g.
signup_completed,booking_created,subscription_started. - Agree on naming: consistent spelling (e.g. object_action), fixed parameters, no personal data in event names or parameters.
- Implement across platforms: the same events on iOS, Android and web, ideally through a shared tracking layer in the code.
- Test and document: check events in a debug view before release and maintain the plan as a living document.
Twenty to forty well-chosen events are plenty for most apps. More data does not automatically mean more insight – but it does mean more maintenance and more privacy risk.
Tools: from Firebase to self-hosting
The right tool depends on privacy requirements, budget and analytical needs. Broadly there are three groups:
- Platform analytics: solutions such as Google Analytics for Firebase are quick to integrate and work well with crash reporting, remote config and experiments. The data sits with a US provider, however, which calls for a careful privacy assessment.
- Product analytics: tools such as Amplitude, Mixpanel or PostHog specialise in funnels, cohorts and user paths. Some offer EU hosting or self-hosting.
- Privacy-focused options: self-hosted systems such as Matomo or your own event pipeline in your infrastructure give you full control of the data, but require more effort to set up and run.
The built-in statistics in App Store Connect and Google Play Console add store conversion, installs and ratings to the picture. For technical quality, dedicated monitoring and crash reporting is worthwhile.
Privacy: consent, ATT and data minimisation
Under the German TDDDG, which implements the EU ePrivacy rules, accessing information on a user’s device generally requires consent unless it is strictly necessary. For most analytics solutions this means collecting data only after consent, unless a careful legal review concludes otherwise. Our article on tracking consent in apps goes into detail. This is not legal advice.
On iOS, App Tracking Transparency applies as well: if you track users across other companies’ apps or websites, you must ask for permission via the system prompt first. You also need to declare data use truthfully in the privacy sections of both stores. Practical principles:
- Collect only what you need for a specific question.
- Use pseudonymous IDs instead of e-mail addresses or names.
- Define retention periods and delete raw data regularly.
- Expect reduced data volumes because not every user consents, and interpret metrics as a sample accordingly.
Common app analytics mistakes
- Averages instead of cohorts: a rising average can hide growth in new installs while retention is falling.
- No versioning: without the app version in events, you cannot isolate the effects of a release.
- No reconciliation with backend data: revenue and bookings should come from the system that actually processes them; client events are for behavioural analysis.
- Interpreting results without experiments: controlled A/B tests in apps show whether a change works more reliably than a before-and-after comparison.
Conclusion
Good app analytics starts with a small set of decision-relevant KPIs along the user lifecycle, a clear north star metric and a clean tracking plan. The tool matters less, as long as privacy, consent and data quality are right. If you want analytics planned properly from day one in a new app, we include the tracking plan in our fixed-price offer.
Frequently Asked Questions
Which KPI matters most for a new app?
Early on, activation rate and day-7 retention are especially telling, because they show whether users experience the core value and come back. Revenue metrics only become reliable once this foundation is in place.
Do I need consent for app analytics?
In many cases yes, because analytics SDKs access information on the device, which under EU ePrivacy rules generally requires consent. Whether an exemption applies should be assessed legally on a case-by-case basis.
How many events should an app track?
Twenty to forty carefully defined events are enough for most apps. What matters is that each event serves a specific question and is implemented identically on every platform.
What is the difference between ARPU and LTV?
ARPU is the average revenue per user over a period, such as a month. LTV estimates the total revenue a user generates over their entire lifetime with the app and therefore depends heavily on retention.