Using Studio Recommendations: apply, reject, log, and apply-as-variant

Review and Apply AI Recommendations in Studio

Studio Recommendations help you improve existing flow messages faster by suggesting concrete changes.
This guide shows how to review recommendations, apply them one-by-one or in bulk, create variants, reject irrelevant suggestions, and use the activity log.

Before you start

Open your flow in Studio.
In the top bar, click Recommendations.
If recommendations exist, you’ll see a badge count on the button.

How Recommendations work

Each recommendation targets a specific message in your flow.
When you hover or click a recommendation, Studio focuses the matching node so you can review context before making changes.

You can then choose one of four actions:

Apply: accept this single recommendation and update the flow.
Apply all: apply all currently listed recommendations in one pass.
Apply as variant: add the recommendation as a variant instead of replacing the original.
Reject: remove the recommendation from the active list and mark it as rejected.

Recommended review workflow

Use this order to avoid accidental bulk edits:

Open Recommendations and scan all items.
Hover/click each item to inspect the target node in the canvas.
Start with Apply on high-confidence changes.
Use Apply as variant when you want to test copy alternatives.
Use Reject for low-quality or out-of-context suggestions.
Use Apply all only after you’ve reviewed remaining suggestions.

Action details

Apply one recommendation

Click Apply on a card.
Studio updates the flow and logs the decision as accepted.

Apply all recommendations

Click Apply all at the top of the sidebar.
Studio attempts to apply all pending items.
Use this when most suggestions are already reviewed and low-risk.

Apply as variant

Click Apply as variant on a recommendation.
Studio converts that action into a variant-style addition (instead of a hard replace).
Best for experimentation, A/B-like copy testing, or preserving current wording.

Reject a recommendation

Click Reject.
The recommendation is removed from the active queue and logged as rejected.
Use this to keep your queue focused and reduce future noise.

Using the Recommendation log

Click Log in the recommendations panel.
Review accepted/rejected history.
Use this as a lightweight audit trail for what changed and what was intentionally skipped.

Best practices

Review recommendations in context, not in isolation.
Prefer Apply as variant for changes that are promising but uncertain.
Reject aggressively when a suggestion does not match campaign intent.
After a batch of applies, do a quick canvas pass and then publish only when satisfied.
Keep your recommendation queue clean so future reviews stay fast.

Troubleshooting

“No recommendations available”