From b71ca84605ee0a34f5e8ac97dced1fa92e797564 Mon Sep 17 00:00:00 2001 From: pwseg Date: Fri, 20 Jun 2025 15:09:27 -0500 Subject: [PATCH 1/3] init randomized splits --- .../event-triggered-journeys-steps.md | 59 +++++++++++++++++++ 1 file changed, 59 insertions(+) diff --git a/src/engage/journeys/event-triggered-journeys-steps.md b/src/engage/journeys/event-triggered-journeys-steps.md index 1aa5799f03..7405b2569e 100644 --- a/src/engage/journeys/event-triggered-journeys-steps.md +++ b/src/engage/journeys/event-triggered-journeys-steps.md @@ -173,6 +173,65 @@ You can use a Data split to branch profiles based on traits or audience membersh This setup helps tailor journey experiences using reliable, preexisting data. Because the Data split step evaluates conditions instantly, it works best with traits or audience membership that Segment has already computed before the profile enters the step. +## Randomized Split + +The **Randomized Split** step lets you experiment with and test the performance of different journey paths. You can add up to five branches, assign each one a percentage, and Segment will randomly send users down one of the branches based on the configured distribution. + +This step is useful for A/B testing, holdout groups, and comparing different channels or messaging strategies within a single journey. + +For example, you might create a randomized split that sends 40% of users to an email campaign, 40% to an SMS campaign, and 20% to a control group. Once users move through the split, you can evaluate which approach performed best. + +### How randomized split works + +When a profile reaches the Randomized Split step: + +1. Segment randomly assigns the profile to one of the branches based on the defined percentages. +2. The profile immediately moves down the assigned path. +3. By default, if a user re-enters the journey later, they’re assigned a new random branch. You can optionally choose to keep them in the same branch each time they re-enter. + +Each journey instance (epoch) is evaluated independently. This means a user could be assigned to different branches across multiple entries, unless you enable consistent assignment. + +### Configuration options + +You can configure a Randomized Split step with the following options: + +| Setting | Description | +|---------------------------------|-----------------------------------------------------------------------------| +| Branches | Add up to five branches. Each branch must be assigned a percentage. | +| Distribution percentages | Define what portion of users should go down each branch. Total must be 100%. | +| Branch naming | Branches are labeled alphabetically (e.g., Branch A, Branch B). | +| Consistent branch assignment | Optionally ensure a user always enters the same branch on re-entry. | + +Segment blocks publishing if the percentages don’t add up to 100%, or if any percentage is left blank. + +> info "Branch assignment is random" +> The Randomized Split step uses probabilistic logic to assign users to branches. At lower volumes, actual distribution may not exactly match your configured percentages, but it tends to even out at scale. + +### Add a Randomized Split + +To add a Randomized Split step: + +1. From the journey canvas, click **+** to add a new step. +2. Select **Randomized Split**. +3. Give the step a name. +4. Add up to five branches and assign a percentage to each one. +5. (Optional) Enable **Keep branch assignment consistent** if you want users to always go down the same branch on re-entry. +6. Click **Save**. + +Once configured, Segment routes profiles through this step based on your distribution settings. + +### Analyze performance + +After users pass through the Randomized Split step, you can view historical and in-progress counts for each branch in the **Journey Overview**. + +You can measure results by: + +- Total journey instances (epochs) +- Unique profiles +- Funnel view or in-progress view + +This helps you evaluate which branch is performing best and informs how you might structure future journeys. + ## Send to Destination The **Send to Destination** step lets you send journey data to one of your [configured Engage destinations](/docs/connections/destinations/), enabling real-time integration with tools like marketing platforms, analytics systems, or custom endpoints. From 89289f97cec4ba50bd778e7ae9f4b305c99469c4 Mon Sep 17 00:00:00 2001 From: pwseg Date: Fri, 20 Jun 2025 15:10:18 -0500 Subject: [PATCH 2/3] clarification --- .../journeys/event-triggered-journeys-steps.md | 18 +++++++++--------- 1 file changed, 9 insertions(+), 9 deletions(-) diff --git a/src/engage/journeys/event-triggered-journeys-steps.md b/src/engage/journeys/event-triggered-journeys-steps.md index 7405b2569e..2fce2b9757 100644 --- a/src/engage/journeys/event-triggered-journeys-steps.md +++ b/src/engage/journeys/event-triggered-journeys-steps.md @@ -173,7 +173,7 @@ You can use a Data split to branch profiles based on traits or audience membersh This setup helps tailor journey experiences using reliable, preexisting data. Because the Data split step evaluates conditions instantly, it works best with traits or audience membership that Segment has already computed before the profile enters the step. -## Randomized Split +## Randomized Split (V2) The **Randomized Split** step lets you experiment with and test the performance of different journey paths. You can add up to five branches, assign each one a percentage, and Segment will randomly send users down one of the branches based on the configured distribution. @@ -181,7 +181,7 @@ This step is useful for A/B testing, holdout groups, and comparing different cha For example, you might create a randomized split that sends 40% of users to an email campaign, 40% to an SMS campaign, and 20% to a control group. Once users move through the split, you can evaluate which approach performed best. -### How randomized split works +### How Randomized Split works When a profile reaches the Randomized Split step: @@ -189,18 +189,18 @@ When a profile reaches the Randomized Split step: 2. The profile immediately moves down the assigned path. 3. By default, if a user re-enters the journey later, they’re assigned a new random branch. You can optionally choose to keep them in the same branch each time they re-enter. -Each journey instance (epoch) is evaluated independently. This means a user could be assigned to different branches across multiple entries, unless you enable consistent assignment. +Segment evaluates each journey instance independently. This means a user could be assigned to different branches across multiple entries, unless you enable consistent assignment. ### Configuration options You can configure a Randomized Split step with the following options: -| Setting | Description | -|---------------------------------|-----------------------------------------------------------------------------| -| Branches | Add up to five branches. Each branch must be assigned a percentage. | -| Distribution percentages | Define what portion of users should go down each branch. Total must be 100%. | -| Branch naming | Branches are labeled alphabetically (e.g., Branch A, Branch B). | -| Consistent branch assignment | Optionally ensure a user always enters the same branch on re-entry. | +| Setting | Description | +| ---------------------------- | ---------------------------------------------------------------------------- | +| Branches | Add up to five branches. Each branch must be assigned a percentage. | +| Distribution percentages | Define what portion of users should go down each branch. Total must be 100%. | +| Branch naming | Branches are labeled alphabetically (for example, Branch A, Branch B). | +| Consistent branch assignment | Optionally ensure a user always enters the same branch on re-entry. | Segment blocks publishing if the percentages don’t add up to 100%, or if any percentage is left blank. From 0807ee710156d6bd126fe2a27ce1fd6762c5370d Mon Sep 17 00:00:00 2001 From: pwseg Date: Fri, 20 Jun 2025 15:12:34 -0500 Subject: [PATCH 3/3] last cleanup [netlify-build] --- .../journeys/event-triggered-journeys-steps.md | 16 +++++----------- 1 file changed, 5 insertions(+), 11 deletions(-) diff --git a/src/engage/journeys/event-triggered-journeys-steps.md b/src/engage/journeys/event-triggered-journeys-steps.md index 2fce2b9757..853a262440 100644 --- a/src/engage/journeys/event-triggered-journeys-steps.md +++ b/src/engage/journeys/event-triggered-journeys-steps.md @@ -202,18 +202,16 @@ You can configure a Randomized Split step with the following options: | Branch naming | Branches are labeled alphabetically (for example, Branch A, Branch B). | | Consistent branch assignment | Optionally ensure a user always enters the same branch on re-entry. | -Segment blocks publishing if the percentages don’t add up to 100%, or if any percentage is left blank. +Segment won't publish your journey if the percentages don’t add up to 100%, or if any percentage is left blank. > info "Branch assignment is random" > The Randomized Split step uses probabilistic logic to assign users to branches. At lower volumes, actual distribution may not exactly match your configured percentages, but it tends to even out at scale. -### Add a Randomized Split - -To add a Randomized Split step: +To add a Randomized Split to your journey: 1. From the journey canvas, click **+** to add a new step. 2. Select **Randomized Split**. -3. Give the step a name. +3. Give the step a unique name. 4. Add up to five branches and assign a percentage to each one. 5. (Optional) Enable **Keep branch assignment consistent** if you want users to always go down the same branch on re-entry. 6. Click **Save**. @@ -222,13 +220,9 @@ Once configured, Segment routes profiles through this step based on your distrib ### Analyze performance -After users pass through the Randomized Split step, you can view historical and in-progress counts for each branch in the **Journey Overview**. - -You can measure results by: +After users pass through the Randomized Split step, you can view historical and in-progress counts for each branch in the Journey Overview. -- Total journey instances (epochs) -- Unique profiles -- Funnel view or in-progress view +You can measure results by total journey instances, unique profiles, funnel view, and in-progress view. This helps you evaluate which branch is performing best and informs how you might structure future journeys.