Skip to content

How to Measure Ecommerce Retention

By Lucas Sahli, Founder, MezzoriPublished Last updated

TL;DR: Measure retention at both the message level and the customer level. Revenue per recipient and attributed orders describe message response; repeat-purchase cohorts show whether customers return over time. To estimate incremental lift, use a valid controlled test rather than treating attributed revenue as causal proof.

Which metrics belong in a retention report?

Layer Metric What it helps you understand
Message response Clicks, placed orders, revenue per recipient Response under the reporting rules
Customer behavior Repeat-purchase rate, time to second order Whether and when customers return
List health Bounces, unsubscribes, complaints Whether the program respects the audience
Business context Total store revenue, discounts, returns Whether message performance fits the business

How do you define repeat-purchase rate?

Choose a customer cohort and a fixed observation period. Count the customers who make a second order within that window and divide by the eligible customers in the cohort. Use the same treatment of cancellations and returns for each comparison. Recently acquired customers need enough time before they can be compared fairly.

The repeat-purchase metrics guide gives a worked example.

What does revenue per recipient mean?

It is attributed revenue divided by recipients for the selected send or reporting period. It helps compare messages with different audience sizes, but audiences, products and promotions still affect the comparison. A tiny sample can produce a misleading result.

Why are open rates not enough?

Privacy features can create opens that do not reflect someone reading the email. Use clicks and purchase behavior alongside delivery and list-health signals. Do not choose a winner from open rate alone.

Is attributed revenue additional revenue?

No. Attribution assigns orders to messages according to configured rules and windows. A customer might have ordered anyway, and email and SMS reports can overlap. Do not add channel figures together without checking how orders are counted.

See Klaviyo attribution and incrementality.

How can a controlled test help?

Randomly assign eligible customers to treatment and holdout groups where the setup supports it. Define the outcome and period before sending. Compare the groups while accounting for uncertainty, sample size and contamination from other messages. If a valid test is not possible, describe the result as a trend or attributed performance.

How should the report guide the next action?

Explain what changed, which audience received it and which limitations apply. Pick a specific next question rather than making several changes at once. Our retention optimization service uses this approach; the Klaviyo email marketing guide covers the broader setup.

Frequently asked questions

Is a larger email revenue share always better?

No. Share can rise because other channels or total store sales fell. Review the full business context and customer outcomes.

Can we compare email and SMS directly?

Only with context. Channel costs, consented audiences and attribution rules differ, and the same order may appear in both reports.

What if we do not have enough customers for a test?

Avoid calling small differences conclusive. Fix clear technical issues and collect more data before estimating lift.

How often should we review retention?

Review message and list-health issues regularly, and customer cohorts over a period that fits your purchase cycle.

More guides on retention measurement

Related topic

Give your customers a reason to come back.

Tell us about your store and your current email and SMS setup. Let’s find the right next step for your retention program.