AnalysisAdvertising

The second order changes the story.

A campaign’s first sale is only the opening number. Compare what its customers actually contribute over the same period.

AI-generated monochrome photograph of stacked glass sorting trays with metal tokens, a return channel and a blank ledger on a dark desk.
SAME CUSTOMER AGE / COMPLETE ECONOMICSAI-GENERATED EDITORIAL ARTWORK
THE USEFUL PART

Compare mature cohorts at the same age. Keep observed contribution, forecast lifetime value, and advertising incrementality as separate questions.

The first order is an opening scene

A campaign produces large baskets and an attractive revenue-to-spend ratio. Another brings in smaller first orders. It is tempting to move the budget immediately, especially when the presentation has room for only one number.

Then the next few weeks happen. Some purchases come back. A discount attracts people who never order at the normal price. Another group buys a refill without another acquisition campaign. Delivery costs differ. Customer support spends more time explaining a promise the ad made a little too confidently.

The original ranking may survive. It may also reverse. The useful question is what each acquisition group contributes over the same observed period, after the costs and adjustments connected to its orders. This is not permission to explain away a losing campaign with imaginary lifetime value. It is a reason to keep measuring what actually happens after checkout.

Put customers on the same clock

A cohort is a group defined by an entry event. Here, use each customer's first paid order, then attach the acquisition label used by your reporting. Shopify's customer cohort reports likewise organize customers around first-order timing and offer measures including retention and sales. Those reports provide a way to describe behavior; they do not establish the causal effect of an advertising channel. [1]

Compare day thirty with day thirty. A September customer has had much longer to reorder than somebody acquired yesterday. Put an incomplete future cell on the report as unavailable, not zero. Zero means you observed no activity during a completed interval. Unavailable means you have not observed that interval yet.

For a fair ninety-day summary, either wait until every included customer has ninety days of follow-up or use an explicitly defined mature subset. Report how many people that leaves. A rolling report that silently drops the newest customers can also change its composition, so keep the eligibility rule visible.

Use stable customer and order identifiers. Email changes, guest checkout, merges, and cross-device activity can make one person look like several. Resolve what the available data supports and label what it cannot. Perfect identity resolution should not be smuggled in as a spreadsheet assumption.

FIELD NOTES / COMPARE EQUAL CUSTOMER AGE
FOLLOW-UP AVAILABLEDAY 0DAY 30DAY 60DAY 90
Mature groupObservedObservedObservedObserved
Newer groupObservedObservedNot yet observedNot yet observed
Recent groupObservedNot yet observedNot yet observedNot yet observed

A future cell is unavailable, not zero. Only the mature group can supply a complete ninety-day result here.

Conceptual observation grid, not customer data. Each dot indicates an available follow-up point; it does not encode revenue or retention.

Watch the apparent winner change

Here is a hypothetical comparison of two one-hundred-customer cohorts. Each required $5,000 in acquisition media. Every customer has a complete ninety-day follow-up period. Values are invented and exclude sales taxes.

Cohort A begins with $12,500 of initial merchandise revenue after its initial discounts. B begins with $10,000. Divide those amounts by acquisition spend and the opening ratios are 2.5× and 2.0×. This is a custom revenue-to-spend comparison, not necessarily the number either ad platform reports.

Ninety-day measureCohort ACohort B
Merchandise sales before discounts and reversals$15,000$14,000
Discounts−$1,500−$500
Sales reversals and refunds in this example−$2,500−$500
Net merchandise sales$11,000$13,000
Reconciled product and variable operating costs−$6,000−$6,000
Contribution before acquisition media$5,000$7,000
Acquisition media−$5,000−$5,000
Contribution after acquisition media$0$2,000

The initially larger revenue result leaves no contribution after acquisition at this horizon. B leaves $20 per original customer. Neither result is company profit: fixed overhead and any omitted costs still exist. Nor is it lifetime value. It is observed contribution over ninety days under the specified assumptions.

The $6,000 cost line includes product, variable fulfillment, payment, and return-handling costs and is already reconciled for recoverable inventory. If a returned item can be sold again, do not blindly subtract its full cost twice. Keep the adjustment trail attached to the order.

FIELD NOTES / THE OPENING RANKING CAN REVERSE
COHORT A / 100 CUSTOMERS2.5×

initial revenue / media

$0

ninety-day contribution
after acquisition media

COHORT B / 100 CUSTOMERS2.0×

initial revenue / media

$2,000

ninety-day contribution
after acquisition media

Invented cohorts with $5,000 acquisition media each. Initial ratios are custom revenue-to-spend measures. Ninety-day contribution includes the adjustments and reconciled variable costs in the table; fixed overhead is excluded.

Reconcile the money before explaining the people

Shopify currently defines net sales as gross sales minus discounts and sales reversals. Reversals are broader than a simple count of returned parcels. Product profit reporting also depends on the recorded item costs and is not automatically contribution after every variable operating expense. [2] [3]

That distinction matters when assembling your own report. A payment refund, a return event, and a reduction in a sales measure may describe related activity without being interchangeable rows to subtract. Link the adjustment once, retain its reason and date, and reconcile totals against the source systems.

Daily reports can place the original sale in one period and its reversal in another. A cohort report follows the original acquisition group across those periods. Shopify's documentation on reporting discrepancies explains that differences in dates and reporting logic can produce different totals. A mismatch is something to trace, not smooth over with a convenient balancing entry. [4]

Start with a small set of orders you can follow by hand: an ordinary order, a partial refund, a returned item restored to inventory, a canceled order, and a repeat purchase. If those do not reconcile, a larger dashboard will produce the same confusion more impressively.

Investigate the promise and the product mix

Avoid turning this into a moral ranking of customers. People can return a product because the ad suggested the wrong fit, the size guide was poor, delivery missed an important date, or the item arrived damaged. A campaign can inherit those problems without causing them.

Separate useful dimensions before telling a story: product, introductory offer, fulfillment region, stock status, acquisition period, and the promise made in the creative. Do not slice the data into dozens of tiny groups until one supports the answer you wanted. Choose a plausible explanation, inspect the relevant records, and design a follow-up comparison.

Acquisition labels deserve the same restraint. A profitable group tagged to a channel does not tell you how many of those customers would have arrived anyway. Cohort economics and advertising incrementality answer different questions. Both can matter to the spending decision.

Set a budget rule that can survive next month

Agree which horizon informs the decision and which early measures act as warnings. A business selling monthly refills has a different observation problem from one selling a sofa. The horizon should fit the repeat behavior and the cash the business can afford to wait for.

Keep forecasts separate from observed results. If you model future repeat orders, display the assumptions beside the mature data and test them against later cohorts. Do not quietly extend the time horizon every time a campaign misses its target.

A practical review might examine first-order contribution, thirty-day reversals, and ninety-day contribution together. A change in any one of them prompts a specific investigation. The goal is a budget decision that improves as customers have time to become customers, rather than one that freezes at the most flattering screenshot of checkout.

PUT IT TO WORK / INTERACTIVE TOOLWork through your ad economicsEditable assumptions. Clear calculations. Save your scenario.

Sources & notes

Primary references for the product details and research discussed above. The examples and practical recommendations are our editorial interpretation.

  1. Shopify Help CenterCustomers reports and customer cohort analysis ↗

    Describes first-order cohorts, retention and sales measures, and available filters. These are descriptive reports, not causal campaign experiments.

  2. Shopify Help CenterSales reports ↗

    Current net-sales definition subtracts discounts and sales reversals from gross sales. Reported sales, collected cash, and profit are different measures.

  3. Shopify Help CenterProfit reports ↗

    Explains dependence on recorded item costs and the treatment of product profit. It does not supply every variable operating or acquisition cost.

  4. Shopify Help CenterSales discrepancies in Shopify Analytics ↗

    Explains timing and reporting differences, including how sales and payments reports treat adjustments.

CHECKED / 18 NOVEMBER 2026

About this article

Prepared by JDGMNT Editorial on 18 November 2026. Both customer cohorts and all financial values are invented. Each has one hundred customers with ninety complete days of follow-up. Costs are reconciled for recoverable inventory; contribution is not company profit, lifetime value, or an incrementality estimate. References are dated to the source check above. The artwork was generated for this collection.

← BACK TO RESEARCH

JDGMNT / INDEX

Find a useful idea.

JDGMNT / CONTACT

What could be more accessible?

Tell us about a question, an access barrier, or a possible connection to our work.

Your message goes directly to the JDGMNT team. Please don’t include sensitive personal information.

EXPLORE JDGMNT

MORE KNOWLEDGE. MORE POSSIBILITY.
MAKING ACCESS POSSIBLE.