Automation Settings

Automation settings control how much Shuttlebase does on its own and how much waits for your explicit decision. There are three settings. Changing them requires the change account settings permission. See Users and permissions.

The three settings

Setting	Default	What it controls
Auto-launch approved candidates	On	Whether approved experiments launch without a further decision from you.
Auto-end losing experiments	On	Whether Shuttlebase ends a losing experiment on its own.
Auto-deploy winning experiments	Off	Whether Shuttlebase rolls out a winning variant to all traffic on its own.

Auto-launch approved candidates

When this is on, an approved experiment joins the launch queue and the scheduler launches it automatically. You still approve every experiment before it can launch. This setting only removes the extra step between approval and launch. See Reviewing and approving for the approval flow.

Auto-end losing experiments

When this is on, Shuttlebase ends an experiment once the recommendations engine calls it a loser. Recommendations only fire after minimum runtimes and conversion counts are met, so experiments are not ended on early noise. When this is off, Shuttlebase sends you an end request instead and the experiment keeps running until you respond.

Auto-deploy winning experiments

When this is on, Shuttlebase rolls out a winning variant to 100% of traffic once the recommendations engine calls the win. When this is off, which is the default, Shuttlebase sends you a deploy request and waits for your approval. Deploying changes what every visitor sees, so most accounts keep this decision manual. See Experiment terminal states.

Max concurrent experiments

Every account has a cap on how many experiments can run at the same time. The cap protects result quality by limiting how much of your traffic is split across tests at once.

The launch scheduler works within this cap:

• It launches approved experiments in priority order (high, medium, low).
• It never launches past the cap. When the account is full, approved experiments wait with the status Awaiting Capacity and launch as slots open.
• It avoids collisions. Two experiments that change the same pages do not run at once. The later one waits with the status Blocked by Live Experiment.

There is no calendar for scheduling launches yourself. The scheduler handles timing. If you want an experiment held back, ask Shuttlebase’s team to put it On Hold. See Experiment lifecycle for all statuses.