Runner AI
DocumentationAutomate, test, and create

Experiments

Choose a store goal and traffic allocation, launch Experiments, and follow experiment cycles and proposed changes.

Experiments let Runner test controlled storefront variations against a selected business goal.

Entry path

Open a store project → MoreExperiments.

Before you start

  • Publish the store and confirm it can receive real traffic.
  • Choose one primary goal that matches the store’s current priority.
  • Choose a traffic allocation you are comfortable assigning to experiment variations.
  • Understand that launch starts an experiment process; it does not guarantee a performance gain.

Launch controls and click results

ControlWhat happens after you select it
Goal cardSelects the metric Runner should optimize, such as conversion, add to cart, checkout initiated, banner engagement, button click-through, or average order value.
Traffic slider or presetSets the share of eligible traffic assigned to the experiment within the allowed range.
Launch ExperimentsAuthorizes and starts the experiment for the selected store, goal, and traffic value.
Experiment title or View ExperimentOpens the experiment detail workspace.
All Experiments, Running, or StoppedFilters the experiment list by status.

Start an experiment

  1. Open Experiments.
  2. If the publication wall appears, publish and verify the store first.
  3. Read every goal and select the one tied to the decision you need.
  4. Set the traffic allocation.
  5. Review the store, goal, and traffic one more time.
  6. Select Launch Experiments once.
  7. Wait for the experiment card to appear.
  8. Open the experiment to follow planning, changes, cycles, and measurements.

Experiment-card controls

Control or sectionWhat it shows or does
Status badgeShows Pending, Running, Completed, Failed, or Cancelled. The Stopped list filter groups completed, failed, and cancelled experiments.
Cycle and TrafficShows the current cycle count and allocation.
PreviewShows the experiment variation when a preview is available.
ChangesLists storefront changes associated with the latest cycle.
AI InsightOpens a pending suggestion such as a traffic adjustment.
AllowApproves the displayed suggestion with its current value.
CancelDismisses that pending suggestion.
More-actions menuOpens Stop and Apply Changes when those actions are available.

Troubleshooting

StateMeaning and next step
Publish wallThe store has no successful live deployment. Publish before enabling Experiments.
Waitlist or upgrade screenThe feature or plan is not available to this workspace.
Preparing or planningRunner is setting up the next cycle. Wait for a plan or visible request.
Waiting for approvalRunner needs a decision before it can continue that suggestion. Read it before choosing Allow.
No changes yetThe current experiment has not produced a visible variation change.
FailedOpen the experiment and preserve the visible error for support before starting a replacement.