This methodology explains how the E-commerce Statistics hub defines metrics, standardizes benchmarks, and selects sources so numbers remain comparable across pages.
Use it when you need to understand scope, definitions, and why two reports can publish different results for the same metric.
Back to the hub:
E-commerce Statistics.
You may also want to open:
Sources
and Glossary.
How we choose sources
The goal is cite-ready benchmarks: clear scope, clear definition, and stable references.
- Primary sources first: official reports, reputable research publishers, and original datasets.
- Clear scope required: market, segment, geography, and time window must be stated or inferable from the source.
- Definition clarity: we avoid using numbers when the definition of the metric is ambiguous (e.g., “share” not defined).
- Cross-checking: when feasible, we include a second reputable source as “range framing” to show reasonable variance.
- Stable URLs preferred: reports and permanent pages are favored over short-lived announcements.
If you see a “range” on a dataset page, it exists to prevent a false impression of precision. When in doubt, cite the primary source list on that dataset page.
Standardization rules
What we standardize to keep benchmark comparisons valid.
Metric definitions
- CR (conversion rate): orders ÷ sessions × 100 (unless the source explicitly uses a different denominator).
- AOV: revenue ÷ orders (scope must be consistent: include/exclude tax and shipping consistently).
- Abandonment: (carts − orders) ÷ carts × 100, noting what “cart” means in the source.
- Share metrics: share of transaction value vs share of transactions are not interchangeable.
Currencies and units
- We keep the source currency when citing (USD vs GBP vs PLN) to avoid conversion ambiguity.
- We keep the source units (bn vs billion; T vs trillion) and label them clearly.
- No inflation adjustments unless the source does it explicitly.
Online vs total retail scope
- Retail e-commerce sales is treated as a retail scope metric.
- Business e-commerce sales can include large B2B volumes and should not be mixed with retail e-commerce.
- Returns: online returns rate is typically higher than overall retail returns rate; we label both explicitly.
Forecast vs actual
We publish forecasts because they are frequently cited, but we keep them clearly labeled.
- Actual: measured historical values reported by the source.
- Forecast: forward-looking estimates, often shown as a time series through a common horizon (e.g., 2028).
- Rule: dataset tables should make it obvious which years are actual vs forecast.
- Rule: if a page includes forecasts, the source and forecast publication date should be included in Sources on that page.
Forecast horizons (e.g., 2028) are included because researchers and journalists frequently cite them as “expected by YEAR” statements.
Quality notes
How to interpret uncertainty and variance across datasets.
- Different panels produce different results: merchant panels, analytics panels, and survey panels can disagree.
- Category mix changes outcomes: returns and AOV are highly category dependent.
- Device definitions vary: some sources merge tablet into mobile; others separate it.
- “Share” requires a denominator: value share vs transaction share matters especially for payments.
How to cite dataset pages
The simplest citation rule: cite the dataset page, not only the hub.
- Link to the specific dataset page you used (e.g., cart abandonment, AOV, market size).
- Include the year/time window as shown on that dataset page.
- Include scope language (e.g., “retail e-commerce sales worldwide”, “share of transaction value”).
- Use the primary source listed on the dataset page when you need original-report citations.
Browse datasets by silo on the hub:
E-commerce Statistics.