The Census Bureau does not classify a business as e-commerce based on email use. Email coordination of contracts and purchase orders is how nearly every modern business operates and does not, by itself, make a company an e-commerce seller. The Census Bureau's definition turns on whether goods or services are ordered through an internet-based system β and the agency's own FAQ answers the email question directly: orders placed via email are not counted as e-commerce.
That single distinction trips up plenty of small-business owners, accountants, and even software vendors who build "e-commerce" features that never touch a shopping cart. Here's how the definition actually works, where the lines fall, and why getting it right matters for surveys, taxes, and competitive analysis.
The Census Bureau's Working Definition Turns on the Order Channel
The U.S. Census Bureau defines e-commerce as the sale of goods or services where the order is placed over an internet-based system. In its Current Business Surveys and the Annual Retail Trade Survey, the agency asks respondents to report the dollar value of shipments or sales conducted through e-commerce, and that figure covers orders accepted online, over EDI (electronic data interchange) networks, and through extranets.
The key word is ordered. The Census Bureau's measurement framework treats e-commerce as a transaction method, not a technology stack. A business can run a fully paperless back office with email, cloud accounting, and digital signatures and still report zero e-commerce sales if customers place orders by phone, mail, or face-to-face conversation.
According to the Census Bureau's own e-commerce FAQ, the definition covers orders placed via the internet, EDI, or extranet, and explicitly excludes orders placed by phone, fax, or email. That exclusion is stated in the agency's guidance rather than inferred, which makes it the cleanest test available to a business owner asking "am I e-commerce?"
The definition also fixes the point of measurement at order placement, not fulfillment. Payment method doesn't decide it either. A customer who orders through a web form and pays by mailed check is an e-commerce order; a customer who emails a signed PDF contract and pays by wire is not. That asymmetry surprises people, but it's how the survey instrument is built.
Email Contracts and Purchase Orders Fail the Census Test
Take the scenario head-on: a wholesaler emails a price sheet, negotiates terms over a thread, attaches a purchase order as a PDF, gets a countersigned contract back, and invoices through an accounting platform. Every step is electronic, and none of it counts as e-commerce under the Census Bureau's definition. The order was transmitted by email, and email sits on the excluded list.
This is the most common false positive among business owners. The intuition runs "we do everything online, so we must be e-commerce." The Census Bureau's framing runs "was the order placed through an internet-based ordering system, EDI, or extranet?" An emailed PDF is a document delivery channel, not an ordering system in the survey's sense.
The distinction holds even when the email thread includes a clickable payment link. If a customer replies to an emailed invoice with payment via an online portal, the sales value is attributed to whatever channel originated the order β and that channel was email. The payment rail doesn't retroactively convert the order.
Business-to-business firms feel this most acutely. Forrester and other analysts have long noted that B2B commerce lags B2C in digital order capture because relationship-driven sales run on email and phone. The Census Bureau's definition simply reflects that reality: a company can be 100% digital in its operations and still sit outside the e-commerce universe as measured.
Three Transactions That Test the Boundary
The clearest way to internalize the definition is to run transactions through it. Consider a furniture maker that sells through three channels.
Web store order. A customer adds a chair to a cart on the maker's site, checks out, and pays by card. Order placed through an internet-based system β counts as e-commerce.
EDI order from a retailer. A national retail chain sends a purchase order through its EDI network. The maker's system accepts it automatically. The Census Bureau counts EDI orders as e-commerce, which is why large B2B suppliers often report substantial e-commerce revenue with no consumer-facing storefront.
Emailed purchase order from a hotel. A hotel's procurement manager emails a PO for 40 chairs. The maker countersigns and invoices. Under the Census Bureau's guidance, this is not e-commerce β the order arrived by email.
The middle case is the one people miss. The Census Bureau explicitly includes EDI in its e-commerce definition, so a business with no website can still be an e-commerce business if it accepts orders through EDI. Conversely, a business with a beautiful website whose customers mostly call in orders may report a surprisingly small e-commerce share.
| Transaction | Order channel | Census e-commerce? | Why |
|---|---|---|---|
| Web cart checkout | Internet-based system | Yes | Order placed online |
| EDI purchase order | EDI network | Yes | EDI explicitly included |
| Emailed PDF purchase order | No | Email explicitly excluded | |
| Phone order with online payment link | Phone | No | Order originated by phone |
| Extranet portal order | Extranet | Yes | Extranet explicitly included |
| Faxed signed contract | Fax | No | Fax excluded |
If your revenue is material and your mix looks like that table, the classification question has real consequences for how you complete Census surveys.
Why the Distinction Matters Beyond Survey Paperwork
For most small businesses, misclassification is a survey-accuracy problem, not a legal one. Census Bureau economic surveys β the Annual Retail Trade Survey, the Annual Wholesale Trade Survey, and the economic census β ask respondents to break out e-commerce sales. An owner who lumps emailed orders into the e-commerce line inflates the agency's e-commerce totals and muddies the industry benchmarks everyone else relies on.
The downstream effects show up in published statistics. The Census Bureau's Quarterly Retail E-Commerce Sales report β the headline number financial press quotes each quarter β only counts retail e-commerce sales as orders placed through the internet, EDI, or extranet. When the agency reported that e-commerce accounted for roughly 16% of total retail sales in recent quarters (the exact share moves each quarter, so confirm the latest figure on the Census Bureau's retail e-commerce page), that percentage is built on the narrow definition. Emailed orders sit in the brick-and-mortar denominator, not the numerator.
Tax treatment is a separate regime entirely. State sales tax nexus rules, marketplace facilitator laws, and economic nexus thresholds are set by state legislatures and revenue departments, not by the Census Bureau, and several states do treat emailed orders differently from web orders for sourcing and sourcing-related purposes. Check with your state's department of revenue or a qualified tax professional before assuming the Census definition maps onto your filing obligations.
Industry benchmarking has the same gap. If you're comparing your "e-commerce share of revenue" to an industry average pulled from Census data, and you've counted emailed B2B orders as e-commerce, your number isn't comparable. The benchmark measures order-channel origin; your internal metric may measure something broader.
What Counts as an Internet-Based Ordering System
The Census Bureau doesn't publish a granular technical spec for "internet-based system," but its survey guidance and FAQ give enough to draw the line. An internet-based ordering system is one where the customer places the order through a networked interface β a website, a mobile app, an online portal, an extranet, or an EDI connection β without email serving as the transmission method for the order itself.
That framing has aged reasonably well. When the Census Bureau first built its e-commerce measures in the late 1990s, the categories were websites and EDI. The extranet category was added as B2B portals proliferated. Email was excluded from the start, and the agency has kept it excluded through every revision of the survey instruments, which is why the exclusion shows up in the FAQ language today.
Application programming interfaces complicate the picture at the margins. If a customer's procurement system pushes an order directly into your order management system through an API β no human typing an email β that looks much more like EDI than like email, and the ordering channel is the integrated system, not a message. The Census Bureau's published guidance doesn't carve out APIs specifically, so document the channel and, if the value is material, confirm treatment with the agency's economic survey staff.
A useful operational test: can the order be accepted, validated, and entered into your systems without a human reading an email? If yes, you're likely inside the e-commerce definition. If a person has to open a message, download an attachment, and key the order in, you're outside it.
Pairing Email With a Real Ordering Portal Flips the Answer
Plenty of B2B companies run a hybrid model: email drives the relationship, and a portal captures the order. That configuration is e-commerce under the Census definition, and it's the single most common way skeptical owners discover they've been under-reporting.
Picture a packaging supplier that quotes by email, negotiates by phone, and then directs the customer to a logged-in portal to place the order and track shipments. The email is the sales conversation. The portal is the order channel. Census counts the portal order. The supplier's e-commerce revenue is real even though no consumer would call the business an online store.
The reverse arrangement produces the opposite result. A company that runs a public product catalog online but takes all orders by emailed PO is not e-commerce by the Census definition, no matter how sophisticated the catalog. The catalog is marketing; the order channel is email. Owners routinely assume a website plus a product list equals e-commerce, and the survey question quietly disagrees.
If you want to move revenue into the e-commerce bucket β for benchmarking accuracy or because a customer or investor expects it β the lever is order capture, not marketing spend. Route orders through a portal, an EDI connection, or an online checkout and the classification follows. Leave order capture in email and it doesn't, regardless of how digital everything around it looks.
One practical consideration: portal adoption among B2B buyers is uneven. Many procurement teams still prefer email because it leaves an auditable thread in their own systems. A supplier that builds a portal and finds half its customers still emailing POs will report a split e-commerce share β portal orders in, emailed orders out β which is exactly what the Census Bureau's definition produces.
The Measurement History Shows Why Email Was Excluded
The Census Bureau began formal e-commerce measurement in the late 1990s, when the agency added e-commerce questions to its economic surveys amid the first wave of dot-com retail. The early survey instruments drew a line between "orders received via the internet" and everything else, and EDI was folded in because large B2B suppliers had been ordering electronically for years before the web existed.
Email landed on the excluded side for a practical reason. An emailed order is not machine-readable in the way a website checkout or an EDI transaction is, and the agency needed a definition that respondents could apply consistently across industries. EDI has standardized message formats; a website order has a structured record; an email is free text plus an attachment. The exclusion wasn't a judgment about how modern email commerce is. It was a measurement decision about what could be counted reliably.
The definition has held stable across survey revisions since, which is unusual in federal statistics. The agency's FAQ language on phone, fax, and email exclusions has persisted because the underlying logic hasn't changed: the survey measures order-channel origin, and email remains a document exchange medium rather than an ordering system.
That stability is useful for business owners. You don't have to guess which year's definition applies. The current guidance matches the guidance from a decade ago on the email question, and the agency's published methodology pages document the EDI and extranet inclusions consistently.
How Retail and Wholesale Surveys Apply the Definition Differently
The same definition produces different-looking questions depending on which Census survey you're answering. The Annual Retail Trade Survey asks retailers to report e-commerce sales as a share of total sales, and the reported figures feed the quarterly retail e-commerce release. The Annual Wholesale Trade Survey asks a parallel question for wholesalers, where EDI and extranet orders dominate the e-commerce total.
The wholesale case is instructive. Wholesale e-commerce in Census data looks nothing like consumer e-commerce β there's no Prime delivery, no cart abandonment. It's EDI purchase orders flowing between ERP systems and B2B portals where buyers reorder pallets of product. If a wholesaler uses email to coordinate the terms and then receives the actual order through EDI, the order counts, and that pattern is common enough that wholesale e-commerce totals run high relative to the consumer-facing numbers most people picture.
For manufacturers, the economic census asks about e-commerce shipments rather than sales, capturing orders received through internet-based systems, EDI, and extranets. A contract manufacturer that takes emailed POs from three large customers and EDI orders from two others reports only the EDI share as e-commerce, even if the email customers are larger.
This survey-to-survey variation matters if you fill out more than one form. The core definition is consistent, but the reporting unit differs: sales for retail and wholesale, shipments for manufacturing. Match your internal bookkeeping to the survey you're completing rather than assuming one number fits all.
Common Misreadings That Produce Wrong Answers
The most frequent error is equating "digital business" with "e-commerce business." A law firm that runs entirely on email, cloud document management, and e-signature is a digital business and a zero e-commerce business under the Census definition. The two concepts measure different things.
The second error is treating payment method as the classifier. A customer who orders by phone and pays through an online gateway has not placed an e-commerce order, and a customer who orders through a web form and mails a check has. The order channel decides; the payment rail doesn't.
The third error is assuming a website automatically creates e-commerce. A brochure site, a catalog without a checkout, and a contact form that generates an email all leave the order channel in email or phone territory. Only when the site accepts the order does the classification flip.
A fourth, subtler misreading: some owners assume all B2B transactions are excluded. The Census Bureau's inclusion of EDI and extranets exists precisely because B2B is heavily represented in e-commerce totals, even when no consumer-facing checkout exists. A supplier that takes EDI orders from a big-box retailer is an e-commerce business by the agency's measure.
| Misreading | What it claims | Correct Census treatment |
|---|---|---|
| "We're fully digital, so we're e-commerce" | Email-run businesses count | Order channel decides, not back-office tech |
| "Online payment equals e-commerce" | Payment method classifies | Order placement classifies |
| "Having a website equals e-commerce" | Brochure sites count | Only sites that accept orders count |
| "All B2B is excluded" | No B2B e-commerce | EDI and extranet orders count |
| "Emailed POs with online invoicing count" | Payment link converts order | Order originated in email, so excluded |
If any row describes your assumptions, the fix is a channel audit: pull a sample of last quarter's orders, tag each by how it arrived, and total the portal, web, and EDI share separately from the email and phone share. That's the number the survey is asking for.
FAQ
Is an emailed purchase order ever counted as e-commerce by the Census Bureau?
No. The Census Bureau's e-commerce FAQ explicitly excludes orders placed by email, along with phone and fax. The order must arrive through an internet-based ordering system, EDI, or an extranet to count. If your customer emails a signed PO and you key it into your system, that revenue belongs in your non-e-commerce total.
Does my business count as e-commerce if I have a website but take orders by email?
Not under the Census definition. A website that displays products but doesn't accept the order leaves the order channel in email territory. The classification flips only when customers place orders through the site, a portal, an app, or an EDI connection. Marketing presence and order capture are separate questions in the survey.
How does the Census Bureau treat EDI orders compared with web orders?
Both count as e-commerce. The agency includes EDI networks and extranets alongside internet-based ordering systems in its definition. This is why large B2B suppliers often report high e-commerce shares with no consumer storefront. The distinction between EDI and web orders doesn't affect the yes/no classification, only the channel mix within the e-commerce total.
Do I need to report e-commerce separately on Census surveys if my share is small?
Follow the survey instructions for your form. Retail, wholesale, and manufacturing surveys each ask for e-commerce values as a component of total sales or shipments, and the reporting unit differs by survey. If your e-commerce share is genuinely zero because all orders arrive by email or phone, report it as zero rather than estimating a figure.
Is the Census e-commerce definition the same one state tax departments use?
No. State sales tax nexus and sourcing rules are set by legislatures and revenue departments, and several states treat email-originated orders differently from portal orders for sourcing purposes. The Census definition is a statistical measurement standard, not a tax rule. Confirm treatment with your state's department of revenue or a qualified tax professional.
Why did the Census Bureau exclude email when email is clearly electronic?
Because the survey measures the order channel, and an email is a document exchange rather than a structured ordering system. EDI has standardized formats and web checkouts produce structured records; free-text email with attachments doesn't. The exclusion is a measurement decision that has stayed consistent through survey revisions, not a claim that email commerce is less modern.
What's the practical way to audit my own e-commerce share?
Pull a representative sample of orders from the last quarter and tag each by arrival channel: web checkout, app, portal, EDI, email, phone, or in person. Total the first four categories against overall revenue. That ratio approximates what the Census survey wants, and it gives you a defensible number for benchmarking against published industry figures.
If I add an ordering portal, does my historical classification change?
Only prospectively. Orders placed through the portal from the launch date forward count as e-commerce; orders that arrived by email before the portal existed remain outside the definition. When you complete a survey covering a period that spans the change, report the split based on when each order was placed and how it arrived.
This article is general educational information about how the U.S. Census Bureau defines and measures e-commerce. It is not legal, tax, or accounting advice, and it makes no promises about survey outcomes or tax treatment. Definitions applied by state revenue departments, the IRS, and other agencies differ from the Census statistical standard. Consult a qualified tax professional or your state's department of revenue before relying on this distinction for filing or compliance decisions.