Published by Take Rate · Definitions reviewed 9 November 2026
Quick answer
Read what was counted before comparing the result. A recovery percentage, revenue per successful delivery and net sales answer different questions. Match the population, period, attribution rule and money basis, then check what refunds and follow-up can still change.
Six definitions worth keeping separate
These are specific published definitions and a comparison rule, not a census of recovery platforms.
Recovered-cart rate
Recovered carts ÷ all abandoned carts ×100, within the declared recovery-program window.
Do not compare it with a sent-email, delivered-message or eligible-checkout denominator.
Start with the worked example below, then record two public claims. Mark each check from their actual definitions. The worksheet does not fetch links or inspect accounts.
Worked arithmetic is available below. Enable JavaScript to change and export the comparison.
The same revenue, a different denominator
Worked claim A:300÷100 successful deliveries=3.00 per delivery. Worked claim B:300÷200=1.50 per delivery. These are authored inputs, not merchant results.
Even matching definitions do not establish a fair performance comparison. Audience, product mix, season and allocation can still differ. Attribution does not establish what would have happened without the message.
Use the supplied worked inputs, then replace them with your own declared record.
Worked result
Inside the declared window
The conversion is 72.00 hours after the anchor, against a 120-hour window.
Elapsed hours
72
Declared window, hours
120
Window setting basis
Worked assumption
Eligibility under this comparison
Within the declared window only
This compares one declared anchor with one conversion. It does not reproduce multi-touch attribution, bot filters or a provider’s account history. A historical unknown remains unknown.
Inputs stay in this page. The tool makes no account, store or provider requests and does not save your entries.
Use these worked amounts to see why the label matters. They are arithmetic examples, not a merchant dataset.
Gross product sales
1,000
Less discounts
100
Less sales-reversal magnitude
150
Net sales
750
Add shipping 50 and tax 80, with no fees
Total sales 880
Deductions are positive magnitudes in this worked calculation:1,000−100−150=750. Shopify reports sales reversals as negative values. When working from signed columns, do not subtract a negative reversal a second time. Refunded shipping and taxes have their own columns.
A return can appear in a later report period than the original sale. A calendar-period total and a mature original-order cohort can therefore answer different questions.
The comparison worksheet calculates only the declared ratio or amount. A recovered-cart percentage is recovered carts÷abandoned carts×100. Revenue per successful delivery is attributed revenue÷successful deliveries. Net sales and attributed-revenue totals are displayed without dividing them by a message count.
Different definitions keep the claims separate. A missing definition needs clarification. A matching declaration does not verify a provider’s records, equal populations or a causal effect.
The time-window checker subtracts the declared timestamps and compares the elapsed duration. It does not reproduce a complete attribution model. Both timestamps need time zones, historical unknowns stay unknown and exact-boundary inclusion remains a provider question.
Klaviyo says attribution-setting changes can recalculate historical data, and delayed events can update credit. Keep the settings/export date beside an old-period figure. Do not treat a report snapshot as a permanent fact.
For incrementality, Klaviyo’s global-holdout documentation describes excluding eligible profiles from marketing, with transactional messages and overrides treated separately. Its 400,000-profile eligibility and three-month recommendation belong to that product feature; they are not universal requirements for every valid experiment. This page runs no experiment and estimates no lift.
Public sources were checked 9 November 2026. Provider definitions and report versions can change. The source definitions, worksheet rules and worked cases are kept separate so a changed provider fact does not silently become a merchant outcome.
Sources & scope
These links supply definitions, not customer proof or platform performance rankings.
Shopify’s legacy documentation says a checkout can be recovered after an email when the shopper completes checkout independently. The newer automation has a different reporting route. Keep the version and named report explicit rather than treating every recovered label as the same attribution rule.
Discuss your recovery setup
Tell Take Rate what you want to compare and the work you want to discuss.