How to Choose the Primary Metric (Winning Metric) for Your Experiment

When you create a new experiment in GemX, you are required to select a Winning metric. This setting directly affects how GemX evaluates performance, calculates the probability to win, and determines which version is recommended as the winner.

This guide explains how the primary metric works, when to choose Conversion rate vs. Revenue, and how to interpret the result messages shown in your experiment dashboard.

Run Smarter A/B Testing for Your Shopify Store

GemX empowers Shopify merchants to test page variations, optimize funnels, and boost revenue lift.

Start Free Trial

What Is a Winning Metric in GemX

The winning metric is the main performance indicator GemX uses to evaluate your experiment statistically.

It determines:

• Which variant has a higher probability to win

• The percentage shown as “X% probability to win”

• Whether the system recommends a version as the winner

You select the primary metric during experiment setup. Once the experiment starts running, all statistical comparisons are based on that selected metric.

In the experiment overview page, you can always see the selected metric in the top summary bar under Winning metric.

Learn more: Understanding Key Metrics and Session Views in GemX

Available Winning Metrics in GemX

Currently, you can choose one of the following:

Conversion Rate

Conversion rate measures the percentage of visitors who complete a purchase.

This metric is best suited when you are testing:

• Page layout changes

• CTA copy or button placement

• Product descriptions or messaging

• Trust elements and social proof

• UX improvements to reduce friction

Choose Conversion rate when your goal is to improve conversion efficiency and you want to understand which version turns more visitors into buyers.

Important note: Conversion rate focuses purely on behavioral performance. It does not account for order value.

Revenue

Revenue measures the total revenue generated by each variant.

This metric is best suited when you are testing:

• Pricing strategies

• Discount offers

• Bundles or upsells

• Shipping thresholds

• Promotions that may affect order value

Choose Revenue when your primary goal is business impact rather than conversion efficiency alone.

One variant can have a lower conversion rate but generate higher revenue due to a higher average order value.

Which Metric Determines the Probability to Win

Only the selected winning metric is used to calculate:

• Probability to win

• Statistical comparison

• Winner recommendation

For example:

If your primary metric is Conversion rate, then the probability percentage shown for each variant is calculated based on conversion rate only.

The probability to win of each version is displayed in the Performance detail box

The statistical engine evaluates probability using the primary metric only.

Why You See Other Metrics Appear in the Summary Box

GemX displays additional metrics such as:

• Conversion rate

• Average order value

• Revenue per visitor

These metrics help you understand overall business impact beyond a single KPI.

For example:

Variant A may have:

• Higher conversion rate

• Higher AOV

• Higher revenue per visitor

In that case, the system may display that Version A wins on all those metrics.

However, the statistical confidence and probability percentage still rely only on the selected winning metric.

Supporting metrics are shown for context. They help you make informed business decisions, but they are not used for statistical winner calculation unless selected as the primary metric.

What If Your Metrics Conflict

It is common to see scenarios such as:

• Variant A has a higher conversion rate

• Variant B generates higher revenue

In this case:

• If Conversion rate is your primary metric, Variant A may have a higher probability to win

• If Revenue is your primary metric, Variant B may be statistically favored

This is why selecting the correct primary metric at the beginning is important. It ensures the experiment aligns with your business objective.

Best Tips to Choose the Right Primary Metric

• Define your business goal before launching the experiment

• Choose Conversion rate for behavioral optimization

• Choose Revenue for financial impact optimization

• Do not switch evaluation logic mid-test

• Use supporting metrics for context, not to override statistical results

A well-chosen primary metric ensures your experiment results align with your real growth objective.

Learn more: How to Apply the Winning Variation and End Your Test Safely

Need more hands-on support?

Book an onboarding call to get guided setup and expert help.

Book A Call

FAQs

Can I evaluate other metrics like AOV?

Yes. You can review all available metrics in the View all metrics section.

However, GemX only calculates the probability to win for the selected primary metric. Metrics such as AOV, Orders, or Revenue per visitor are shown for business context and must be interpreted manually.

Can I change the primary metric after launching the experiment?

Yes. You can change the winning metric even after the experiment is live.

To update it:

• Go to your experiment list
• Click on the experiment you want to update
• Select Edit experiment
• Update the winning metric and save your changes

Once updated, GemX will recalculate probability to win based on the newly selected primary metric.

Should I always choose Revenue as the primary metric?

Not necessarily.

Choose Conversion rate when optimizing user behavior and funnel efficiency. Choose Revenue when measuring direct financial impact. The correct choice depends on your business goal for that specific experiment.

Realted Topics: 

Getting Started