▸Example Countly call (the "before")
import ly.count.android.sdk.Countly
import ly.count.android.sdk.CountlyConfig
val config = CountlyConfig(application, "YOUR_APP_KEY", "https://your-countly-server.com")
.setLoggingEnabled(true)
.setIdMode(DeviceIdType.OPEN_UDID)
Countly.sharedInstance().init(config)
Countly.sharedInstance().events().recordEvent(
"purchase",
mapOf("user_id" to userId, "value" to price, "currency" to "USD"),
1
)Most analytics SDKs accept arbitrary event parameters — the path of least resistance for an engineer adds user_id, email, phone, address to events over time. Respectlytics's 5-field event schema (event_name, session_id, timestamp, platform, country) rejects extra fields at the API with a 400 — making PII drift architecturally impossible.
☑Remove Countly cleanly
-
1
Remove the Countly SDK from your build (
Countly/ly.count.android:sdk/countly-sdk-react-native-bridge/countly_flutter) -
2
Remove
Countly.sharedInstance().start(with:)initialisation andrecordEvent(...)call sites -
3
Decide whether you'll keep the Countly server running for other apps, or decommission it as part of the migration
-
4
If you used Countly's crash reporting alongside analytics, plan a separate crash-reporter migration (Sentry / Crashlytics / Bugsnag)
⇋Countly vs Respectlytics — no pii collected
| Countly | Respectlytics | |
|---|---|---|
| Accepts arbitrary event parameters | Yes (per-event params or properties) | No (5-field schema, extras rejected with 400) |
| Persistent user identifier | — see tool note above | No (session_id rotates every 2h, RAM-only) |
| IP address stored | — see tool note above | Used transiently to derive country, then discarded |
| Per-user historical record | — see tool note above | Out of scope (use your account system) |
❓Frequently asked questions
How do we attribute revenue to a specific user without `user_id`?
You don't — at least not in your analytics. Your billing system (Stripe, RevenueCat, App Store Connect) is the authoritative source of per-user revenue, with refund-aware totals and the appropriate access controls. Respectlytics tells you the product signal (conversion rate, funnel completion) at the session level; per-user revenue lives where it belongs.
What if our analytics team needs to slice by user segment (paid vs free)?
Encode the segment into the event name. Instead of track('paywall_view', { tier: 'paid' }), fire track('paywall_view_paid_user') and track('paywall_view_free_user'). The aggregation buckets them automatically; no per-user identity is stored.
Doesn't the country field count as personal data?
Country-only resolution is generally considered the most minimised form of geographic data and is widely accepted in privacy reviews — but consult your legal team to determine your specific situation. Respectlytics derives country from the request IP server-side, then discards the IP before storing the event.
What happens if a teammate accidentally adds an extra field to a `track` call?
The API returns a 400 with the offending field name in the response body. Your integration test fails on the first run that includes the new field, so the regression is caught at PR review — not after months of unnoticed silent PII collection.