A/B Test Metrics

This page defines every metric that can appear in A/B test analytics. For how to read them (cohort vs non-cohort, target metrics, Effect, P-value), see Analyzing Experiments.

Some metrics — predicted and 1M / 3M / 6M variants — are marked Only admin in the column picker. Admin in this context refers to organization-level Administrators or Owner.

ARPPU

Average Proceeds Revenue Per Paying User. Average proceeds from paying users, including subscriptions and non-renewing purchases associated with the A/B-tested paywall.

Cohort metric — calculated by fetching users who were distributed to the relevant variation within the selected dates.

📘
Note
You can analyze this metrics in ARPPU report as well. Use the Experiment Variations filter — go to the report, click the Add filter +, pick Experiment Variations, and choose your variation(s).

ARPU

Average Proceeds Revenue Per User. Average proceeds across all users assigned to the variation, including subscriptions and non-renewing purchases associated with the A/B-tested paywall.

Cohort metric.

📘
Note
Use the Experiment Variations filter on the ARPU or Cumulative LTV chart to drill in.

ARPAS

Average Revenue Per Active Subscriber. Average revenue from active subscribers — paying and free-trial — including subscriptions and non-renewing purchases associated with the A/B-tested paywall.

Cohort metric.

📘
Note
Use the Experiment Variations filter on the ARPAS chart to drill in.

ARPPU 1M

Average Proceeds Revenue Per Paying User after the 1st month — same as ARPPU but calculated over a 1-month window from variation assignment.

Cohort metric. Only admin.

ARPU 1M

Average Proceeds Revenue Per User after the 1st month — same as ARPU but calculated over a 1-month window from variation assignment.

Cohort metric. Only admin.

ARPAS 1M

Average Revenue Per Active Subscriber after the 1st month — same as ARPAS but calculated over a 1-month window from variation assignment.

Cohort metric. Only admin.

View to Action

Conversion rate from viewing a paywall to performing a valuable action — starting a free trial, starting a paid subscription, or making a non-renewing purchase. Refunds don't matter for this metric.

Requires the SDK to send the Paywall Shown event; otherwise the conversion isn't available.

Cohort metric — number of purchase or trial start events divided by Unique Views in the selected dates.

View to Trial

Conversion rate from viewing a paywall to starting a free trial.

Requires Paywall Shown from the SDK.

Cohort metric.

View to Purchase

Conversion rate from viewing a paywall to a first paid event — starting a paid subscription, converting a free trial, or a non-renewing purchase. Refunds don't matter for this metric.

Requires Paywall Shown from the SDK.

Cohort metric.

Trial Conversion

Conversion rate from starting a free trial to a paid subscription. Refunds don't matter for this metric.

Cohort metric — trial conversions divided by trial starts in the selected dates.

Sales

Total amount billed to customers, excluding refunds and excluding revenue not tied to the A/B-tested paywall.

Non-cohort metric — calculated by summing sales in the selected dates.

Proceeds

Estimated revenue you receive after taxes, store commissions, and refunds. Excludes revenue not tied to the A/B-tested paywall.

Non-cohort metric.

Cohort Proceeds

Proceeds attributed to users who entered the variation in the selected dates, summed over their lifetime.

Cohort metric. Compares variations on long-term revenue contribution per cohort, not just calendar revenue.

Overall Proceeds

Estimated revenue after taxes, store commissions, and refunds — regardless of paywall. Includes revenue from non-renewing purchases and subscriptions purchased or reactivated after the A/B test launched.

Recurring revenue from subscriptions purchased before the test started is excluded.

Non-cohort metric.

Refunds

Total refunds tied to the A/B-tested paywall.

Non-cohort metric.

Trials

Number of free-trial start events tied to the A/B-tested paywall.

Non-cohort metric.

Unique Purchases

Number of first paid events — paid subscription starts, free-trial conversions, and non-renewing purchases. Refunds don't reduce the count.

Non-cohort metric.

Purchases

Total paid transactions — initial purchases, renewals, and non-renewing purchases. Excludes zero-price transactions. Refunds don't reduce the count.

Non-cohort metric.

Unique Views

Number of unique users who viewed a variation's paywall.

Requires Paywall Shown from the SDK.

Non-cohort metric.

Views

Number of Paywall Shown events, counting repeats. Requires the event from the SDK.

Non-cohort metric.

Users

Number of users distributed to the variation.

Non-cohort metric.

Predicted metrics

Apphud uses machine learning to predict future revenue per variation based on current revenue trends and user behavior. Predicted metrics require LTV Predictions to be enabled for your app.

All predicted metrics are cohort metrics (fetching users distributed to a variation in the selected dates) and marked Only admin in the column picker.

For the calculation methodology — payback iteration boundaries by subscription duration, why predictions use the maximum iteration in the window rather than calendar days — see Experiment Predictions.

pARPU 1M

Predicted Average Revenue Per User after the 1st month. Uses machine learning to predict the ARPU that might be achieved on the 1st month.

pARPU 3M

Predicted Average Revenue Per User after the 1st three months. Uses machine learning to predict the ARPU that might be achieved on the 1st three months since purchase.

pARPU 6M

Predicted Average Revenue Per User after the 1st six months. Uses machine learning to predict the ARPU that might be achieved on the 1st six months since purchase.

pARPU 1Y

Predicted Average Revenue Per User after the 1st year. Uses machine learning to predict the ARPU that might be achieved on the 1st year since purchase.

pARPPU 1M

Predicted Average Proceeds Revenue Per Paying User after the 1st month. Uses machine learning to predict the ARPPU that might be achieved on the 1st month.

pARPPU 1Y

Predicted Average Proceeds Revenue Per Paying User after the 1st year. Uses machine learning to predict the ARPPU that might be achieved on the 1st year since purchase.

pARPAS 1M

Predicted Average Proceeds Revenue Per Any Subscriber after the 1st month. Uses machine learning to predict the ARPAS that might be achieved on the 1st month.

pARPAS 1Y

Predicted Average Proceeds Revenue Per Any Subscriber after the 1st year. Uses machine learning to predict the ARPAS that might be achieved on the 1st year since purchase.

pProceeds 1M

Predicted Proceeds revenue after the 1st month. Uses machine learning to predict the Proceeds revenue that might be achieved on the 1st month.

pProceeds 3M

Predicted Proceeds revenue after the 1st three months. Uses machine learning to predict the Proceeds revenue that might be achieved on the 1st three months since purchase.

pProceeds 6M

Predicted Proceeds revenue after the 1st six months. Uses machine learning to predict the Proceeds revenue that might be achieved on the 1st six months since purchase.

pProceeds 1Y

Predicted Proceeds revenue after the 1st year. Uses machine learning to predict the Proceeds revenue that might be achieved on the 1st year since purchase.