Experiments

Run A/B tests to optimize pricing, paywalls, onboarding flows, and feature adoption. AppActor handles user assignment, targeting, and variant delivery.

How It Works

1. Create an experiment on the server with variants and targeting rules
2. Request assignment from the SDK — the server assigns the user to a variant
3. Use the variant to show different UI or behavior
4. Track goals to measure which variant performs better

Getting a Variant

iOS

if let assignment = try await AppActor.shared.getExperimentAssignment(
  experimentKey: "paywall_v2"
) {
  // Assignment exists → user is in experiment
  switch assignment.variantKey {
  case "control":
      showOriginalPaywall()
  case "variant_a":
      showNewPaywall()
  default:
      showOriginalPaywall()
  }

  // Optional payload access
  if let showDiscount = assignment.payload["showDiscount"]?.boolValue, showDiscount {
      showDiscountBadge()
  }
} else {
  // User not targeted or experiment not running
  showOriginalPaywall()
}

Android

Coming Soon

Android SDK is currently in development.

Flutter

Coming Soon

Flutter SDK is currently in development.

React Native

Coming Soon

React Native SDK is currently in development.

Experiment Structure

Key Concepts

Concept	Description
Experiment Key	Unique identifier (e.g., "paywall_v2")
Traffic Allocation	Percentage of eligible users included (in basis points, 10000 = 100%)
Targeting	Rules to filter eligible users (same as Remote Config)
Variant	A named group with a weight and payload
Control	The baseline variant (usually current behavior)
Weight	Percentage of experiment traffic for this variant (in basis points)
Payload	Arbitrary JSON data attached to the variant
Sticky Assignment	Once assigned, a user always sees the same variant

REST API

Request Assignment

POST /v1/experiments/paywall_v2/assignments?app_user_id=user_123
Authorization: Bearer pk_live_your_key_here

Query Parameters:

Parameter	Required	Description
app_user_id	Yes	User ID (max 256 characters)
app_version	No	App version for targeting
country	No	ISO 3166-1 alpha-2 code for targeting

Response (Assigned)

{
  "data": {
    "experiment": {
      "id": "exp_abc123",
      "key": "paywall_v2"
    },
    "variant": {
      "id": "var_xyz789",
      "key": "variant_a",
      "valueType": "json",
      "payload": {
        "layout": "new",
        "showDiscount": true
      }
    },
    "inExperiment": true,
    "assignedAt": "2025-01-15T12:00:00.000Z"
  },
  "requestId": "req_abc123"
}

Response (Not Targeted)

{
  "data": {
    "inExperiment": false,
    "reason": "not_targeted"
  },
  "requestId": "req_abc123"
}

Targeting Conditions

Same targeting system as Remote Config:

Condition	Operators	Example
Platform	eq, neq, in, not_in	iOS only
App Version	eq, gt, gte, lt, lte	v2.0.0+
Country	eq, neq, in, not_in	US and CA
Entitlement	has, not_has	Free users only

Goals

Track what you're optimizing for:

Goal Type	Description
conversion	Did the user convert (e.g., subscribe)?
revenue	Revenue generated
retention	User retention after N days
custom	Custom event tracking

Best Practices

1. Always handle the "not in experiment" case — Show default behavior when the user isn't targeted
2. Use the control variant — Always have a control group for comparison
3. Check assignment early — Fetch the assignment before rendering the UI
4. Don't change experiments mid-flight — Let experiments run to completion for valid results

Next Steps

• Apple Search Ads — Attribution tracking