This article explains how to segment your traffic using Advanced Settings in GemX so you can run more reliable, focused A/B tests.
Learn more: For more details about Advanced Settings in GemX, please refer to this article.
What Is Traffic Segmentation
Traffic segmentation lets you control which visitors are included in your experiment. Instead of showing your test to all visitors, you can limit it to specific groups based on:
-
Device
-
Visitor type
-
Traffic source
-
Market and language
This helps you reduce noise in your data, test more accurately, and match experiments to user intent.
Example: If you're testing a mobile layout, run the test only for mobile visitors to avoid mixing user behavior across devices.
Before You Start
Before setting up traffic segmentation, make sure that:
-
You have created an experiment in GemX
-
Your test has at least one Variant
-
You understand your testing goal (what you want to validate)
Pro tip: Always start with a clear hypothesis before applying segmentation. For example: "Mobile visitors drop off on product pages due to layout issues."
How to Segment Your Traffic in GemX
Follow these steps to access the segmentation settings:
Step 1: From your Shopify admin, go to the GemX: CRO & A/B Testing app.
Step 2: Open the Experiment tab and click your draft experiment to open its settings.

To filter all draft experiments, select the "Experiment status" as "Draft."
If you haven't created an experiment yet, click Create new experiment.

Click "Create new experiment" to create a new testing campaign in GemX.
Step 3: On the experiment settings page, scroll to the Advanced Settings section right under the experiment name. From here, configure the filters described below.

Traffic Split
The traffic split controls how visitors are distributed between the Control and the Variant. By default, traffic is split 50/50. Adjust the slider based on your risk level or rollout strategy.

By default, the traffic is split 50-50 in GemX.
Use cases:
-
50/50 split for accurate comparison
-
80/20 split to limit exposure to a new variation
Best practice: Start with a 50/50 split unless you have a specific reason to reduce risk.
Device Type
Device targeting lets you choose which devices are included in your test:
-
Desktop
-
Tablet
-
Mobile

You can select desktop, tablet, mobile, all devices, or two of three options to segment the visitor based on their device.
Default: All devices selected.
Use cases:
-
Test mobile-specific layouts or UX
-
Run desktop-only experiments for checkout optimization
Important note: If your layout differs significantly across devices, run separate experiments for each device type.
Visitor Type
Segment traffic based on whether visitors are:
-
New visitors
-
Returning visitors
-
Both new and returning visitors

The visitor type is selected as both new & returning visitors by default.
Why it matters: These groups behave differently. New visitors are usually still exploring, while returning visitors are more likely to convert.
Use cases:
-
Show trust elements to new visitors
-
Test offers or urgency for returning visitors
Pro tip: Returning visitors often have higher purchase intent, so use this segment to test offers, bundles, or urgency elements.
Traffic Source
Traffic source segmentation lets you filter visitors based on how they arrive at your store:
-
Direct
-
Email
-
Referral
-
Organic social
-
Organic search
-
Paid social
-
Paid search
-
SMS

You can also segment the traffic based on traffic sources.
Why it matters: Different traffic sources have different intent and behavior.
Use cases: Align your test with traffic intent. For example:
-
Paid traffic: test messaging and value proposition
-
Email traffic: test urgency or offers
Best practice: Avoid combining paid and organic traffic in the same test if your goal is to measure performance accurately.
Market & Language
Market and language segmentation lets you target visitors based on your Shopify Markets setup. You can segment traffic by:
-
Country or region
-
Language

Note: This option is only available if Shopify Markets is configured for your store. If not, you'll see a message prompting you to set it up:
| "No domain / language set up yet. Set up in Markets." |
Use cases:
-
Test localized pricing or currency display
-
Optimize messaging for different languages
-
Compare performance across regions
Important note: Always make sure your Variant is localized when targeting a specific market.
Best Practices for Traffic Segmentation
-
Keep segmentation simple to maintain enough traffic volume
-
Align segmentation with your test hypothesis
-
Focus on high-impact segments such as mobile or paid traffic
-
Run separate tests instead of combining too many variables
-
Wait until you have enough data before making decisions
Pro tip: If your test is taking too long to reach significance, check whether your segmentation is too narrow.
Next Steps
After setting up segmentation, you can:
-
Launch your experiment from the Experiment tab
-
Monitor performance in the Analytics tab
-
Use Heatmaps to understand visitor behavior
-
Apply the winning variation once results are conclusive