Flow Control

Flow control blocks allow you to determine how users progress through a Journey based on attributes, behaviours, or experimentation logic. These steps dynamically shape user paths, enabling precise audience targeting and optimisation.

There are three types of flow control components:

• Filter

• Split

• Experiment

---

Filter

Use a filter step to evaluate users based on static or dynamic conditions. This step creates two distinct branches:

• Qualified – users who meet the filter criteria

• Everyone else – users who do not meet the criteria

Example use case

Filter users who:

• are on the “Premium” subscription plan AND

• have triggered the app_opened event within the past 7 days.

These users will continue down the qualified path, while others will be routed to Everyone else.

Configuration

1. User properties

   Pick an existing audience or create one using related tables, events, and traits. (e.g., subscription_plan = "Premium", signup_date > 01-01-2025, Total Order Count > 25, etc.).

2. Journey activity

   Toggle on to filter users based on journey-related activity or dynamic behaviour (e.g., "has opened the app in the last 7 days").

---

Split

A split step segments users into multiple branches based on the value of a selected property. This is commonly used to customise the experience for different user segments.

Configuration

1. Split on: Select a property to evaluate (e.g., rating, region, or driver_status).

2. Define paths: Create one or more conditions to group users into specific paths:

   • Path 1: rating is less than 2

   • Path 2: rating is between 2 and 4

   • Path 3: rating is greater than 4

Users are evaluated against these conditions in the order listed and routed accordingly. Any user who doesn't match a condition will follow the default everyone else path.

Example use case

Split users by satisfaction rating to deliver tailored follow-up experiences:

• <2: Show apology and collect feedback

• 2–4: Offer support tips

• 4 - 5: Prompt referral invitation

Experiment

An experiment step enables randomised testing across multiple paths to measure and optimise user outcomes. This is useful for A/B or multivariate testing.

Example use case

Test two onboarding email sequences by splitting users 40/60. Track which sequence results in more app opens within 7 days, and automatically switch all future users to the winning version after 7 days.

Configuration

1. Variant distribution

   Specify what percentage of users should go down each path (e.g., 40% to Path 1, 60% to Path 2). You can optionally add more paths or a control group.

2. Conversion tracking

   • Conversion event: Select the event that signifies success (e.g., app_opened).

   • Tracking window: Define how long to measure conversions (e.g., 7 days).

   • Auto-switch to winning path: Automatically route future users to the top-performing path after a fixed period (e.g., 12 days).

3. Control group Set a percentage of users who do not receive anything from the experiment. This helps you to analyse the experiment on an absolute basis – how much did each variant do compared to users who did not receive any communications.

The sum total percentages for all variants and control group should add up to 100%.