A Firebase Analytics alternative with no identifiers

Trackless Telemetry and Google Analytics for Firebase, compared

Checked against Google's own documentation on October 26, 2026

The short answer

Google Analytics for Firebase is free, and it is part of a large toolkit: crash reports, performance traces, A/B testing, Remote Config, audiences, BigQuery export, and links to Google Ads. For many apps that breadth is the right call, and Trackless does not try to match it.

Firebase gives every installation of your app an app-instance ID and uses it to count users. On iOS it also collects the vendor identifier (IDFV), or the advertising identifier (IDFA) when your app links AdSupport and the person allows tracking. On Android it collects the Advertising ID by default. Each event is kept as its own record.

Trackless stores no identifier of any kind and no individual events, only aggregate counts per day. It cannot count people, build audiences, or measure ad campaigns. What you get instead is analytics with nothing in it that could single out a person, and nothing to declare as an identifier on your App Store label.

If you need user counts, crash stack traces, or ad attribution, Firebase answers those and Trackless does not. If you need to know which features are used, what is failing, and how slow it is, with no identifiers on the device or on the server, Trackless is built for that.

At a glance

TracklessFirebase
Identifier storedNoneAn app-instance ID for every installation
Device identifiersNone iOS: IDFV, or IDFA with AdSupport and tracking permission. Android: Advertising ID by default
Individual events storedNo, only daily aggregate countsYes; BigQuery export has a row per event with the pseudonymous ID
IP addressNever read; region comes from the device's settingsUsed for approximate location, not logged or stored
Device modelNot collected; device class only (phone, tablet, desktop)Collected (model, brand, and marketing name)
User counts and audiencesNoYes
Conversion funnelsAggregate step countsYes, per user
Errors By name and code, with errors per session and spike alerts, plus an opt-in count of uncaught exceptions by class or name, with no stack traces Crashlytics crash reports with stack traces
Timing p50 / p90 / p99 and a duration histogram per measurement, with threshold alerts; launch time and main-thread hangs if you turn them on Performance Monitoring traces
A/B testing and Remote ConfigNoYes
SDKsSwift (iOS, macOS), Android, webiOS and other Apple platforms, Android, web, Flutter, Unity, C++
PriceFree to 100,000 events a month, then from $9 a monthAnalytics, Crashlytics, Performance Monitoring and A/B Testing are free
How long data is kept13 months, every plan User- and event-level data 2 or 14 months; standard aggregate reports are not affected by that setting

Choose Firebase if

  • You need to count people: daily and monthly active users, new users, retention.
  • You want crash reports with stack traces. Trackless counts errors, and can count uncaught exceptions by class or name if you turn that on, but it keeps no traces and is not a crash reporter.
  • You run A/B tests or ship configuration changes with Remote Config.
  • You advertise and want audiences, attribution, and Google Ads in the same place.
  • You want raw events in BigQuery to query yourself.
  • You ship with Flutter, Unity, or C++. Trackless has no SDK for those.
  • Cost decides it. Firebase's analytics, crash and performance tools are free.

Choose Trackless if

  • You want analytics that collect no device identifier and no advertising identifier, so there is nothing per person to protect, disclose, or delete.
  • You want an App Store label with nothing under Identifiers.
  • You want usage data that never leaves for an advertising company. Trackless shares telemetry with no third party.
  • You care about what is failing and how slow it is: errors per session, p50 / p90 / p99 timing, and email alerts when error rates climb, a measurement crosses a threshold you set, or an app goes silent.
  • You work with a coding agent and want to hand it your app's production numbers as a bounded, readable pack with no personal data in it.

What is different under the hood

For the general case, A hashed ID is still an ID explains what a stored identifier, hashed or not, lets someone link together, and where Trackless's own protection runs out.

The identifiers

The Firebase SDK generates an app-instance ID for each installation and uses it to identify unique installations and compute user metrics (Google: data collection). On iOS it collects the IDFA only if your app links the AdSupport framework and the person grants permission through App Tracking Transparency; otherwise it collects the IDFV. On Android it collects the Advertising ID by default, through a permission the SDK merges into your manifest. You can turn IDFA and IDFV collection off (Firebase: configure data collection), but the app-instance ID is how Firebase counts users.

Trackless sends no identifier at all. A session lives in memory on the device and ends when the app goes to the background; the next launch is a new, unrelated session. Ingest adds each event to a daily count for its name and coarse context (platform, OS version, device class, app version, region, language), and nothing is written per event. Two sessions on the same phone are indistinguishable from two sessions on two phones.

The IP address

Google says Analytics does not log or store IP addresses, and uses them to protect the service and to estimate where in the world your users are (Google: safeguarding your data). Trackless's application code never reads the IP address at all. Region comes from the device's own region setting, which the person controls.

What is kept, and for how long

Firebase keeps events individually. The BigQuery export has a row per event with a pseudonymous user ID, an optional user ID you set, device model, and the advertising or vendor identifier (Google: BigQuery export schema). In the Analytics property you choose to keep user- and event-level data for 2 or 14 months; that setting affects explorations and funnel reports, not the standard aggregate reports (Google: data retention).

Trackless keeps aggregate counts for 13 months, then deletes them automatically. Deleting your account deletes your apps' data after a 7-day grace period.

Errors and timing

Crashlytics groups crash reports into issues by the frames in each stack trace (Firebase: Crashlytics), and Performance Monitoring records custom code traces broken down by country, device, app version, and OS level (Firebase: Performance Monitoring).

Trackless counts handled errors by name and code, reports errors per session, and emails you when a rate spikes. It keeps no stack traces. With captureUncaught on, it also counts each exception no code caught, by its class or name and nothing else; most Swift crashes are signals rather than exceptions, and are not caught. Timing measurements arrive as p50 / p90 / p99 and a duration histogram per name, with alerts when one crosses a threshold you set, and the SDK can measure launch time and main-thread hangs if you turn them on.

The App Store label

Google's disclosure guide lists what the SDK handles: an app-instance ID for every installation, general location derived from masked IP addresses, in-app purchases, and app lifecycle events such as screen views and sessions, plus optional features such as a user ID and Google signals that add more (Google: App Store data disclosure). Its Google Play guide lists the app-instance ID, the Advertising ID, coarse location, and purchases as collected automatically (Google: Play data disclosure).

With Trackless you declare Product Interaction, Other Diagnostic Data, and Performance Data (and Crash Data, if you turn on uncaught-exception counts), none linked to the user, none used for tracking, and nothing under Identifiers. The privacy guide walks through every screen, and the Google Play Data safety answers too.

Handing production to your agent

Your agent gets one app's Agent pack: the aggregate counts for the window you choose, with the instructions that tell an agent how to read them. You can copy or download it from the Agent pack page, or let your agent fetch it itself from Trackless's read-only MCP endpoint with a read token (Connect your agent). Trackless calls no model. Because storage is aggregate-only, a whole app's production fits in a context window, and there is no personal data in the pack, because there is none anywhere. Once your agent has the pack, those counts sit with whichever service it uses, under its terms.

Pricing

Firebase lists Analytics, Crashlytics, Performance Monitoring, and A/B Testing among its no-cost products, with no payment method needed (Firebase: pricing). Trackless cannot beat free.

Trackless Free is 100,000 events a month for 2 apps, with no credit card. Indie is $9 a month for 1,000,000 events and 5 apps. Trackless never upgrades a plan automatically and never charges overage. All plans are on the pricing table.

Moving from Firebase

The calls map closely. Trackless takes one optional detail value per event rather than a dictionary of parameters, because each value becomes its own aggregate count.

Firebase (Swift)Trackless (Swift)
Analytics.logEvent("export", parameters: nil)Trackless.feature("export")
logEvent("export", parameters: ["format": "csv"])Trackless.feature("export", detail: "csv")
logEvent(AnalyticsEventScreenView, …)Trackless.view("settings")
Performance.startTrace(name: "sync") / stop()Trackless.startTimer("sync") / Trackless.stopTimer("sync")
Crashlytics.crashlytics().record(error: error)Trackless.error("import_failed", code: "json_decode")

Analytics.setUserID, user properties, audiences, and anything else that depends on telling installations apart have no equivalent, and Trackless keeps no stack traces: with captureUncaught on, it counts uncaught exceptions by class or name, which is not a crash report. If you keep Crashlytics for crash reports, it keeps collecting what it collects. The documentation covers each platform, and every SDK ships a guide written for coding agents, so you can point yours at it and let it do the migration.

Sources

Every Firebase fact here comes from Google's own pages, linked above, read on October 26, 2026. Products change; if something on this page is out of date, tell us at feedback@tracklesstelemetry.com and we will correct it.