Traffic Isolation

Traffic isolation lets you control what percentage of your total page traffic enters an experiment, reserving the rest for your normal store experience or other experiments. This is useful when you want to limit exposure, run multiple experiments without overlap, or protect revenue on high-traffic pages.

What Is Traffic Isolation?

By default, experiments in Elevate include all eligible visitors — if a visitor lands on a page with an active experiment and matches the audience targeting rules, they're included in the test.

Traffic isolation changes this. When enabled, you specify what percentage of your total traffic should enter the experiment (e.g., 10%). The remaining traffic (90%) either goes to your store as normal or enters other isolated experiments.

For example:

Without isolation: 100% of visitors to the page are split between control and variation
With 10% isolation: Only 10% of visitors enter the experiment. The other 90% always see the original page.

When to Use Traffic Isolation

Limiting risk on high-traffic pages — If your homepage gets 50,000 visitors/day and you're testing a bold change, you might want to start with only 10% of traffic entering the experiment
Running multiple experiments simultaneously — If you have two experiments on the same page, you can isolate each to its own traffic slice to prevent visitors from being in both tests at once
Gradual rollouts — Start with a small percentage, monitor results, then increase the allocation if things look good
Protecting revenue — During critical periods (like BFCM), you might want to limit experiment exposure to reduce risk

How It Works

When traffic isolation is enabled for an experiment, the Elevate script handles the allocation before assigning visitors to variations:

A visitor arrives at a page with active experiments
If any experiment has traffic isolation enabled, the script first determines which experiment (if any) the visitor should enter — based on the isolation percentages
If the visitor is selected for an isolated experiment, they enter that experiment and are assigned a variation normally
If the visitor falls into the remaining (non-isolated) traffic, they enter any non-isolated experiments running on that page
A visitor who's already been assigned to an isolated experiment stays in that experiment on return visits

The key point: traffic isolation happens before variation assignment. It determines whether a visitor enters the experiment at all, not which variation they see once they're in.

How to Enable It

Traffic isolation is available as a toggle during experiment creation. You'll find it below the traffic allocation (variation split) section.

Expand the Traffic Isolation section
Toggle it on
Set your desired percentage using the slider

The slider shows how your total traffic is distributed:

Current Experiment — The percentage allocated to this experiment
Other Experiments — Traffic allocated to other isolated experiments (if any)
Remaining — Traffic that doesn't enter any isolated experiment

The default when first enabled is 10%, which is a safe starting point for most experiments.

Running Multiple Isolated Experiments

When you have multiple experiments with traffic isolation enabled on the same page, they share the 100% traffic pool. The slider shows you how traffic is distributed across all isolated experiments.

For example, if you have:

Experiment A: 10% traffic isolation
Experiment B: 15% traffic isolation
Remaining: 75% goes to non-isolated experiments or normal traffic

Each visitor can only be in one isolated experiment. The allocation is random — a visitor has a 10% chance of entering Experiment A, a 15% chance of entering Experiment B, and a 75% chance of not entering either isolated experiment.

Important: The total traffic allocated across all isolated experiments cannot exceed 100%. The slider automatically caps your allocation based on what's already reserved by other experiments.