Cart recovery measurement kit

By Rene Siller. Sources checked 2026-11-04.

Reconcile eligible groups, revenue and costs. Then test a shopper task against facts and permissions you can actually verify.

## Quick answer

Use comparable eligible groups and one cost definition. This worksheet normalizes the arithmetic; it does not verify randomization or infer statistical confidence from totals. [Read the method.](index.html#method)

Source context: Zettler Digital, measuring email and SMS incrementality (https://zettlerdigital.com/blogs/news/how-to-measure-email-sms-incrementality-with-klaviyo); Common Thread Collective, measurements for profitability (https://commonthreadco.com/blogs/coachs-corner/building-useful-measurements-for-profitability).

Method and downloads

Download method

## Reconcile the group totals.

Use the number of eligible shoppers assigned to each group, including those who did not buy. Use the same observation window, net-revenue definition and order-cost treatment. Check duplicate orders and subscription renewals before entering totals.

Treated group

Control group

Costs not already included in either group’s variable order costs. Enter zero explicitly if none apply.

Point estimate for the treated group after entered program costs

Enter both groups

This scales control contribution to the treated group’s size. Arithmetic does not verify the experiment or establish statistical confidence.

The worksheet will show both group contributions, the scaled control and the cost deduction.

Different group sizes are normalized before subtraction. Rounding occurs only at the final cent.

## Check the comparison before the number.

A group difference needs a suitable design to support a causal interpretation. Confirm how shoppers were assigned, whether groups could receive the other treatment, and whether eligibility or tracking changed during the period.

Contribution per eligible shopperGroup net revenue minus group variable order costs, divided by eligible shoppers. Use the same unit in both groups.Normalize before subtractingMultiply the control contribution per eligible shopper by the treated group count. Subtract that value and additional program costs from treated contribution.What totals leave outTotals do not contain person-level variation, so this worksheet cannot calculate confidence intervals, statistical power or a reliable test duration. A positive estimate alone is insufficient evidence.Count each cost onceRefunds and discounts already deducted from net revenue stay there. Product, payment and fulfillment costs belong in the order-cost definition. Additional program fees go in the program-cost field only if they have not been counted elsewhere.

## Check a shopper conversation.

A helpful-sounding reply can still use the wrong fact or take an unauthorized action. Write the expected behavior from the merchant’s actual policy and permissions, then record what happened. An empty or unrun case is not a pass.

No cases recorded in this session.

## Download the measurement templates

### Group measurement sheet

Empty fields for consistent groups, costs, period and reconciliation notes.

Download group sheet, CSV

### Conversation case sheet

Verified facts, permissions, expected behavior and actual observations.

Download case sheet, CSV

### Cost input sheet

Keep the additional program cost separate from costs already counted.

Download cost sheet, CSV

### Method and source notes

Use the same definitions in the tool, spreadsheet and review.

Download method, Markdown

## Sources and scope

- Zettler Digital, measuring email and SMS incrementality (https://zettlerdigital.com/blogs/news/how-to-measure-email-sms-incrementality-with-klaviyo). A practical group and order-reconciliation method. This worksheet does not inherit a universal test duration or holdout size.
- Common Thread Collective, measurements for profitability (https://commonthreadco.com/blogs/coachs-corner/building-useful-measurements-for-profitability). Revenue, costs and valid experimentation answer different parts of the decision.
- Red Stag, interpreting cart abandonment (https://redstagfulfillment.com/percentage-of-online-shoppers-abandon-their-cart/). Cart counts and attributed orders do not establish recoverable revenue or additional profit.

These sources explain the definitions and measurement principles used here. The worksheet applies the arithmetic to the values you enter.

## Your entries stay in your browser

Amounts and case records are processed in your browser. Your entries are not sent to a server or saved between visits. Export the record to keep it.

## Set up your next conversation case

Use the case sheet to record the shopper task, verified facts, permissions and actual observations.

Download case worksheet

## Publisher disclosure

Zobi publishes this worksheet. Your entries are processed in your browser, not sent to a server. The calculations apply the stated definitions to the values you enter.

Method and calculation definitions updated 4 November 2026. Use the method notes alongside your exported records.

Zobi · zobi.com