E-commerce
May 23, 2026
What percentage of retail sales were online in 2019? In French: what share of retail trade went through online sales in 2019? As with any past year, the answer depends on the country, the statistical series, and the definition of “online”. There is no single fixed global figure that could be cited without a source.
This guide explains how to find and read a percentage for 2019 without making a mistake, why published values may differ depending on the edition of the table (revisions), and how to use 2019 as a reference year with caution, for example before the sharp acceleration of certain digital channels in 2020.
We do not list here a precise “official” percentage for 2019 by region: notices and revisions change; always open the dated source excerpt. For your current management: e-commerce analytics, GA e-commerce tracking, Shopify reports, how e-commerce works.
Three key ideas. First: 2019 often serves as a baseline “before the shock” in economic commentary, which explains why the question comes up in searches. Second: comparing 2019 to 2023 or 2026 without harmonizing definitions gives misleading curves. Third: your store today has no reason to base its digital target on a national aggregate from 2019.
If you're in a hurry: start with the site of your national institute or Eurostat for Europe, download the series that separates online sales in retail, and note the version and scope before copying a percentage into a slide.
Noting the exact URL saves you from searching again in three years during an audit.
For the same issue without the 2019 anchor: see the general logic of online shares in retail (definitions, countries) in your neighboring content on the e-commerce percentage in retail trade.
A tip: when a blog post displays “2019” in the title but cites a 2024 study, check whether the figure is a retrospective reconstruction or a value published at the time; it is not always identical after revision.
Finally, distinguish a full calendar year and a fiscal year if you are matching statistics to your own closings.
Open data bases often have a field for “last updated”; for 2019, a line may have been recalculated years later after additional surveys were incorporated.
If you work in BI software, create a metadata table by series: country, unit, seasonality, access link; you will save time with each new management report.
Data science product teams need to know whether a predictive model uses 2019 as a feature for structure or only as a time dummy; document the choice to avoid information leakage between train and test on long series.
Summary
2019: a benchmark year, not a single answer
Asking for a percentage for 2019 is asking for a historical measure. Institutes still publish it in databases or summary reports, sometimes in the form of indices or shares rather than a single median number.
Why 2019 specifically
Many readers look for a reference point before 2020 to place a later trend or a "normalization" after peaks.
No universal percentage
The "retail" denominator and the "online" numerator vary from country to country; a German figure is not an American figure copied over.
What you actually take away
The method for extracting the data and citing it properly internally.
Students sometimes confuse e-commerce share in retail with year-over-year e-commerce growth; these are different indicators even for the same year.
If your management asks you for "the 2019", ask back "which country and for which exact indicator" before opening ten tabs.
Read a 2019 series without choosing the wrong column
For 2019, as for any year, the methodological note remains the priority. It says which goods are included in retail, how a sale is classified « online », and whether the figures are provisional or final in the edition you are consulting.
Retroactive revisions
A 2019 value published in 2020 may be adjusted in 2022 when new surveys arrive; two « 2019 » PDFs may differ.
Current prices and volume
The value share can move with inflation even if habits are stable; specify the unit: price and margin.
Seasonality
An annual percentage hides peaks; comparing the fourth quarter of 2019 with the fourth quarter of 2018 can tell a different story than the whole year.
Online databases often allow filtering by NACE or product codes; an overly broad « retail » label mixes low-digital items with textiles.
Keep a screenshot or CSV export with the download date; useful when an auditor asks where the slide came from.
Europe: Eurostat, institutes, cautious harmonization
In Europe, Eurostat and national statistical institutes have long published indicators on sales made via the internet. The exact table titles vary; what matters is to identify the row that matches your question and the right country.
Cross-country comparisons
Without a table of equivalence, comparing France and a neighboring country in 2019 may look elegant visually and be methodologically wrong.
UK and others
The geographical and institutional scope of 2019 for the United Kingdom and the EU had a particular political and statistical context; always cross-check the national source.
Report language
Definitions in English and in the local language may use similar terms without being strictly identical.
For a brand that was already delivering across all of Europe in 2019, the dispersion of national shares explains why a “Europe” figure requires a real aggregate and not a hasty average across countries.
BI teams building an internal benchmark can index several countries with the same calendar and the same constant currency to limit exchange-rate artifacts.
United States: retail, e-commerce and revisions
In the United States, retail series and e-commerce components are published by federal agencies; the revisions and merchant categories are the key to interpreting 2019 as well as other years.
Delivery address
In several frameworks, retail e-commerce is analyzed with a destination logic useful for managers, not just for headlines.
Comparison 2019-2020
Public comments on the online share after 2020 often use 2019 as a baseline; keep the same definition for both years if you remake the chart.
Watch the words
“Non-store” or similar expressions do not always strictly cover the same set as “e-commerce” in another table.
US marketplace sellers in 2019 sometimes saw a share of their revenue attributed differently depending on whether they were reading a market study or a public statistic : Amazon, marketplace.
If you are a US scale-up, document how your CRM tagged “web” in 2019 ; stack migrations have erased many internal historical records.
Channels: what “online” already covered in 2019
The store pickup after web order, nascent social commerce, and apps in 2019 were already blurring the “online / store” boundary for the customer, even though the statistics split it cleanly according to internal rules: omnichannel, networks, activate social sales.
Point of Sale
Some brands counted web-store revenue separately; national statistics could classify the same sale differently depending on the recorded order channel.
Pure players
Their apparent share in national retail could be high in a few categories and negligible in other heavy industries.
B2B
Online flows between businesses can be excluded from household retail; always check the denominator.
Hybrid retailers that were already publishing an integrated annual report in 2019 are a useful source for seeing how a large-format store presented its own web share at the time, separate from the macro figures.
For product teams, comparing only photos of 2019 checkouts found on Archive.org can illustrate the UX gap with today without pretending to be a macro-level measure.
Compare 2019 with the following years methodically
Using 2019 as a reference for 2025 or 2026 can help distinguish a “overcorrection” or maturation, provided you do not exclude underlying structural changes (logistics, payments, customer expectations).
Honest comparison baseline
Choose a recent year too for your internal goals; 2019 alone freezes a distant era for a small business founded afterward.
Seasonality and Black Friday
Campaigns at the end of 2019 did not have the same scale as some later periods; compare identical calendar windows.
Pandemic effect
2020 often shifted shoppers to the web temporarily; knowing where the level stabilizes takes several years: benchmarks.
Avoid narratives like “back to 2019” if your internal data show a durably different channel distribution for your audience.
Retail researchers sometimes use fixed-effects models by country; an entrepreneur does not need the formality, but the idea remains useful: control for what is country-specific before comparing two percentages.
For a global brand, a 2019 analysis can still shed light on cultural adoption gaps (payment, postal trust) that persist even after digital acceleration.
Keep an eye on academic work that reconstructs long series; it often clarifies methodological breaks better than journalistic excerpts.
Firms that re-benchmark markets often use 2019 and the latest available year as two reference points; cite both explicitly to avoid hasty conclusions.
Your internal data: does 2019 still exist?
Your business in 2019 may not be comparable to itself today: products, countries, CRM, analytics stack.
Internal data 2019
If you still have exports, check the channel definition; the tags have often been cleaned up since: conversion, improve conversion.
Mobile
The mobile share of traffic in 2019 was already significant for many verticals: mobile first.
Goal 2026
Build a target based on your recent trajectory, not just a memory of 2019: roadmap.
Buyers conducting due diligence on a target may ask whether post-2020 “normalization” returns above or below the 2019 mix; it's a question about the segment, not the national average.
For marketplaces, the 2019 commission files may still exist with the accountant even if the front office has changed four times.
Use 2019 in strategy and communication
Even without publishing a precise 2019 percentage in your strategy, you can use it in an internal memo to explain why you are monitoring certain trend deviations.
Content strategy
Evergreen articles about 2019 attract informational queries; the added value is methodological teaching rather than an isolated figure copied without a source.
Investors
Show that you know where the official data is rather than a round number without a link.
Monitoring
Schedule an annual review of your sources: revisions can slightly shift a 2019 curve.
Legal teams often prefer wording such as “according to [source] edition [date]” rather than a categorical numeric claim without a hyperlink.
If you aggregate multiple countries, note the time zones for the closing of promotional periods at the end of 2019 so you can compare them fairly across regions.
Be wary of recycled 2019 figures without a source
Generalist articles that display a 2019 percentage without a link to the original table should be read with caution. Prefer a direct link to the institute or the database.
Recycled charts
Twitter or LinkedIn visuals age poorly; check the year axis and the legend.
Translations
An imprecise English word (« online ») can mask a narrow definition in the source language.
Consultants and paid studies
Useful for context if the methodology is public; otherwise, label the figure as a sector estimate.
Example : two posts cite « 12 % and 18 % for the same country in 2019 » ; often one includes services, the other only certain goods, or one is in value and the other in share of companies that sold online.
Setting up a team clipboard with the citation template (source, table, date, filter) reduces transcription errors.
Qstomy: avoid random historical stats in customer support
Buyers or readers who discover historical statistics sometimes ask confusing questions to support (“why was this country less digital”). An assistant can direct them to a neutral FAQ on the definitions without quantifying the past.
Qstomy for Shopify: AI chatbot, automated customer support, customer experience.
Links: demo, offers, assisted selling, support, analytics. Client scripts must not invent macro shares by year.
Even without AI, a sentence like “our lead times and availability are those shown on the site today” avoids debating a country ranking in 2019 on chat.
If journalists contact you, pass them to the communications department with the official source rather than to an overworked junior.
Beyond 2019: useful current indicators
Go further into measurement and context, without getting stuck on a single year:
Current view of the channel
Basket, returns, logistics: average order value, returns, inventory.
Acquisition
Today's advertising costs are not those of 2019: marketing costs, traffic.
Loyalty
Long-term value takes precedence over a 2019 snapshot: retention, CAC and LTV.
Internal marketing historians sometimes find specifications from 2019 revealing which digital KPIs were already being tracked; this helps calibrate expectations about what is "comparable" over time.
For a recent scale-up, a reading of cohorts since 2021 is often better than an obsession with 2019 alone.
Summary, FAQ, and Further Reading
In brief
A “online retail 2019” percentage depends on the country and the definition, with no single global value here.
Revisions can shift a 2019 series over time.
2019 often serves as a baseline reference before atypical years, not as a 2026 target for your store.
Always cite source, table, and extraction date.
FAQ
Can you give the exact figure for France in 2019?
Not in this article: open the national institute or Eurostat with the series you are interested in and the table version.
Why do websites contradict each other on 2019?
Revisions, different scopes, or imprecise translations.
Is 2019 still relevant for my business plan?
As a historical benchmark, yes; as the only compass, no: update it with your recent data.
Should I compare 2019 to 2020 without caution?
No: mention the shock and the method; avoid simplistic conclusions.
Does the mobile share in 2019 explain my entire current funnel?
No: UX expectations have changed; test today.
Where can I learn retail statistical vocabulary?
The glossaries of official institutes and a few reliable academic teaching sheets.
For more information
Enzo
May 23, 2026
