Understanding Report Metrics in Convert
Learn how to interpret metrics like Conversions, Confidence, Chance to Win, and Improvement in your Convert experiment reports.
When you're analyzing A/B test results in Convert, clear understanding of each report metric helps you make confident, data-backed decisions. This guide explains the meaning of every key metric shown in your experiment reports—for both Frequentist and Bayesian statistics engines. Each definition is presented in plain language and includes an example for better interpretation.
🔄 Conversions
Definition:
A conversion is counted each time a visitor completes a defined goal (such as a purchase, form submission, or page view) during their participation in an experiment.
Example:
If 100 users see Variation A and 10 of them complete your checkout goal, you have 10 conversions for Variation A.
📈 Improvement
Definition:
Improvement reflects the estimated performance increase (or decrease) of a variation compared to the baseline (original). It’s usually shown as a percentage.
Example:
If Variation B converts at 5% and the baseline converts at 4%, the improvement is +25%.
🧠 Confidence (Frequentist)
Definition:
Confidence tells you how sure we are that the observed improvement is not due to chance. It’s based on statistical significance.
Example:
A 95% confidence means there's only a 5% chance that the improvement you're seeing is just random noise. It's a common threshold for calling a result “statistically significant.”
🎯 Chance to Win (Bayesian)
Definition:
“Chance to Win” is used in Bayesian analysis and shows the probability that a variation is the best performer among all variations.
Example:
If Variation C has a 92% chance to win, it means there’s a 92% likelihood it’s the top variation based on observed data.
📊 Conversion Rate
Definition:
The conversion rate is the percentage of visitors who triggered a goal out of all those who saw a particular variation.
Formula:
Conversions / Visitors × 100
Example:
10 conversions from 200 visitors equals a 5% conversion rate.
👥 Visitors
Definition:
The number of unique people who were exposed to a specific variation in your experiment.
Example:
If 300 users see Variation A and 300 see Variation B, each variation has 300 visitors.
🧮 Total Conversions
Definition:
This shows the sum of all conversions for that variation across all users.
Tip:
Used in multi-goal scenarios or with goals that can fire more than once per user.
🏁 Reached Goal (Boolean Goals)
Definition:
When you're using Boolean goals, this simply shows how many users triggered the goal at least once.
💵 Revenue (if enabled)
Definition:
If you’re using a revenue goal, the report will show the total revenue generated by each variation.
Example:
Variation B leads to 5 purchases totaling $1,000—your revenue for Variation B is $1,000.
🎯 Goal Conversion Rate
Definition:
For each goal, this is the percentage of users who triggered the goal, relative to the number who saw that variation.
Frequentist vs Bayesian Summary:
Metric | Frequentist | Bayesian |
Confidence | ✅ Yes | 🚫 No |
Chance to Win | 🚫 No | ✅ Yes |
Conversion Rate | ✅ Yes | ✅ Yes |
Improvement | ✅ Yes | ✅ Yes |
Conversions | ✅ Yes | ✅ Yes |
If you're unsure which engine you are using, go to Project Settings > Experiment Settings > Statistics Engine.
ℹ️ Note:
Metrics like “Chance to Win” and “Confidence” are calculated differently and should not be directly compared between Bayesian and Frequentist reports.
Need more help interpreting results?
Check out Live Logs for Projects and Experiments to troubleshoot or validate data in real-time.