Before interpreting an online-sales number, identify what was counted, which businesses it covers, when it was measured and how returns or other adjustments were treated. A payment processor's transactions, a retailer's net sales and a national statistical estimate can all be useful while describing different things.
Write a definition beside every number
Create a short definition record: source, geography, seller coverage, channel, period, currency, tax/shipping treatment, return treatment, adjustment method and release date. Mark an unknown field as unknown. If a report leaves a consequential definition unclear, narrow the comparison until it can be resolved.
| Source type | Useful interpretation | Check before comparison |
|---|---|---|
| Merchant report | Sales activity under that business's reporting definitions. | Gross/net basis, cancellations, returns, time zone and order status. |
| Processor or platform report | Transactions or sales within its observed network. | Merchant mix, changing coverage, transaction types and treatment of refunds. |
| National retail estimate | Estimated activity for a defined retail population. | Survey coverage, exclusions, revisions and uncertainty. |
| Consumer survey | What respondents report buying or intending to buy. | Question wording, sample, weighting and intent versus completed purchase. |
The U.S. Census quarterly retail ecommerce report is an example of a national retail series. Its methodology explains that estimates are weighted from a retail sample and that online travel services, financial brokers and ticket agencies are outside its retail scope. It also documents the removal of nonemployers from the series in the April 2025 revision. Use consistent revised data when comparing periods.
Separate merchandise sales from cash movement
Define your own management calculation explicitly, then map it to the report fields. For the example below, gross merchandise sales are before discounts and returns. Net merchandise sales equal gross merchandise sales minus discounts minus returns, with tax and shipping excluded throughout. This is an illustrative definition, not a claim that every platform labels fields this way.
| Measure | Period A | Period B |
|---|---|---|
| Gross merchandise sales | $10,000 | $12,000 |
| Discounts | $500 | $600 |
| Returns deducted in the period | $500 | $900 |
| Net merchandise sales | $9,000 | $10,500 |
| Placed orders counted for this example | 100 | 120 |
| Net sales per placed order | $90.00 | $87.50 |
Gross growth is ($12,000 − $10,000) ÷ $10,000 = 20%. Net growth is ($10,500 − $9,000) ÷ $9,000 = 16.67%, rounded. Net sales per placed order fell from $90 to $87.50 even though total net sales rose. More orders and lower net sales per order can occur together.
These returns are booked in each reporting period; they may relate to earlier purchases. To study the eventual return rate of a set of orders, follow that order cohort through its return window instead. Specify whether your average order metric includes canceled orders and how it handles later refunds.
Settlement deposits introduce further differences such as processing fees, timing, reserves and chargebacks. Reconcile them to the appropriate payment report. An unexplained difference between deposits and merchandise sales is a question to investigate, rather than evidence that one total is wrong.
Check the denominator and the calendar
Growth percent = (new value − old value) ÷ old value × 100. The old value is the denominator. When it is zero, this percentage calculation is undefined; report the absolute change and explain the starting point.
A share moving from 10% to 12% rises by 2 percentage points, or 20% relative to its old share. Label which change you mean. For category shares, also check that the categories use the same total and that overlapping categories have not been added together.
Match period length and time zone. A five-day event, a calendar month and a retail week can differ in their weekday composition and promotional timing. Compare equivalent windows where the question calls for it, and state whether calendar effects have been adjusted statistically.
Census notes that its seasonally adjusted ecommerce estimates account for seasonal, holiday and trading-day variation, while remaining unadjusted for price changes. Seasonal adjustment and inflation adjustment solve different problems.
Distinguish revenue growth from quantity growth
Revenue can rise because more units sell, prices change or customers buy a different mix of items. A higher average order value may reflect larger baskets or a shift toward expensive products. Inspect units, items per order and a consistent product mix before attributing the change to one cause.
Keep the unadjusted figures and the adjustment method together. If quantities, prices or mix cannot be separated with the available data, state that limitation and avoid assigning a precise volume-growth explanation.
Write a defensible interpretation
A useful conclusion names the measure, comparison and limitation: “For these two fictional periods, net merchandise sales increased 16.67% while net sales per placed order declined to $87.50. Returns were recorded when processed, so this comparison does not establish the eventual return rate of either order cohort.”
For a published report, save the release date and version. Preliminary estimates can be revised. Check uncertainty intervals before treating a small change as a clear trend, and compare against the same series rather than choosing a new source that produces a more striking number.
Then connect the result to a question you can act on: whether a promotion improved contribution, whether returns increased for a particular product, or whether staffing matched order volume. A large total becomes useful when its definition supports the decision.