Assign variants for AI decisioning
Wire Content Library AI variants into a journey Decisioning node so GrowthLoop can allocate traffic and optimize toward your metric.
After you generate AI variants in Content Library, assign each variant to a branch under a journey Decisioning node. GrowthLoop allocates traffic across those branches and shifts toward better performers based on the metric you select.
Prerequisites
- A journey ready to edit — see Build a journey
- At least two creative options in Content Library (typically a parent template plus AI variants, or multiple variants)
- A metric that measures the outcome you want to maximize (for example clicks, conversions, or revenue)
- Destinations configured for each branch (GrowthLoop Email/SMS or a composable destination)
How it works
- You generate variants in Content Library.
- You add a Decisioning node and choose an optimization Metric.
- You add a Variation branch for each creative arm.
- Under each Variation, you add a Destination node and assign that arm’s template or variant.
- After the journey runs, View performance insights shows reward rate, uplift, and traffic share by arm.
GrowthLoop uses dynamic allocation across variations—you do not set manual percentages on Decisioning arms the way you do on a classic Experiment node.
Step-by-step guide
Add a Decisioning node
- Open your journey canvas.
- Drag a Decisioning node onto the canvas and connect it from the prior step (for example after an audience entry or delay).
[Screenshot: Decisioning node on the journey canvas]
- On the Decisioning node, open the Metric selector.
- Select the metric to optimize. If you do not have one yet, click Create a metric to maximize and follow Create a metric.
[Screenshot: Decisioning node Metric selector]
Add Variation branches
- Use the Decisioning node’s add-child control to create a Variation branch.
- Repeat until you have one Variation per creative arm you want to test.
- Optionally rename each Variation so your team can tell arms apart (for example “Variant A – urgency CTA”).
[Screenshot: Decisioning node with multiple Variation children]
Assign a variant to each arm
- Under the first Variation, add a Destination node (GrowthLoop Email/SMS or a composable destination such as Iterable or Salesforce Marketing Cloud).
- Configure the destination export as usual — see Trigger campaigns in journeys.
- Assign creative using the path that matches your destination:
GrowthLoop Email or GrowthLoop SMS
For each Variation’s Destination node, open the content picker and select the Content Library variant (or parent template) for that arm. Variants appear with a Variant badge. GrowthLoop sends through SendGrid without creating a separate campaign object per arm, so you can assign Local content on every branch the same way.
[Screenshot: Variation branch Destination node with a Variant template selected]
Composable destinations with campaigns or journeys (Iterable, Salesforce Marketing Cloud)
Keep all decisioning arms inside one platform campaign or journey:
- On the first Variation’s Destination node, pick a Content Library variant (Local content) and confirm with Use template / Publish and use. GrowthLoop uploads the content to the platform and creates the campaign (Iterable) or journey (SFMC), prompting for any required publish settings.
- On each remaining Variation’s Destination node, open the Campaigns tab and select a content variant within that same campaign or journey—do not publish a new Local content item per arm unless you intentionally want a separate campaign/journey for each branch.
NoteUsing Local content on every arm is allowed, but each confirmation creates a new campaign or journey on the platform. The usual pattern is one shared campaign/journey with multiple content variants.
TipKeep everything except the creative constant across arms (same destination type, match fields, and timing) so decisioning can attribute differences to the variant.
Publish and monitor
- Save and publish the journey when configuration is complete.
- After users flow through the Decisioning node, click View performance insights on the node.
- Review per-arm Reward Rate, uplift, and Traffic share. Insights are available after the journey has run.

[Screenshot: Decisioning insights panel showing reward rate and traffic share by arm]
Decisioning vs Experiment nodes
| Decisioning | Experiment | |
|---|---|---|
| Purpose | Automatically optimize toward a metric | Manual A/B (or multivariate) split you define |
| Traffic | Dynamic allocation across arms | Fixed % or count splits you set |
| Best for | AI variants and ongoing creative optimization | Controlled experiments with predetermined splits |
Use Decisioning for Engagement Suite AI variant assignment. Use Experiment when you need a fixed split—see the Experiment node section in Build a journey.
Congrats—you’ve wired AI variants into journey decisioning!
NoteExperiencing any issues? Reach out to us at [email protected] and we'll help resolve the matter.
Updated 3 days ago