An anonymous usage statistics tracking library for iOS using a differential privacy approach.
Beekeeper allows you to get insights about your most important KPIs like daily, weekly or monthly active users, funnels and events and much more without sacrificing your users' privacy.
.package(url: "https://github.com/ChaosCoder/Beekeeper.git", from: "1.0.0")
An event in the app (*e.g. user tapped a button or opened the app) is sent via Beekeeper to your server.
An event, that is fired in your app, includes the following data:
id: String: Random UUID for the event
product: String: The app the event was fired in
timestamp: Date: The precise timestamp the event was fired
name: String: The name of the event
group: String: The group/category of the event
detail: String?: A detail of the event (optional)
value: Double?: A numeric value of the event (optional)
custom: [String]: Custom data
Additionally, each event carries metadata. This metadata is usually the crucial part for the privacy of the user. Let's have a look at the metadata:
previousEvent: String?: Event (name), that was triggered before this event within the same group
previousEventTimestamp: Day?: Date (day precision), when an event of the same, was triggered last time
install: Day: Date (day precision) when the user installed the app
Note: There is no user id, no IP address or other identifying information about the user.
Each event is stored in a database with just the information listed above. Each event is isolated from the history, as the only two links to previous events are too unprecise to be chained together. This way, Beekeeper is anonymous and not only pseudonymous. This property is important for preserving the users' privacy, as pseudonymous solutions can potentially be de-anonymized.
Beekeeper allows you to generate important insights about your app. A basic event that can be tracked in an app is the event of opening the app.
Your daily app sessions can then be calculated by just counting those events on a given day. The daily active users (who could have multiple sessions a day) can be calculated by just counting the events with a
previousEventTimestamp of a past day. The same goes for weekly, monthly or other timeframes.