Online business fraud covers the scams and schemes aimed at a company that sells online, through its checkout, its customer accounts, its refunds or its partners. Older e-commerce fraud lists put payment fraud with stolen cards first and filed it under identity theft. The 2026 merchant survey gives identity theft a row of its own, at 31 percent, and puts refund and policy abuse at the top. A card number works without anyone's identity. A false refund claim needs neither. A fraud detection and prevention program built around stolen cards alone misses the most common type.
What are the most common types of online fraud against businesses in 2026?
Ranked by the share of merchants hit in the 12 months to late 2025 (2026 Global eCommerce Payments & Fraud Report, Figure 24, p. 25):
- Refund and policy abuse, 41 percent. It needs only an account and a claim.
- Friendly fraud, 36 percent. The survey calls it first party misuse: a real customer disputes a real purchase.
- Card testing, 33 percent. Bots try card numbers on cheap checkouts, and merchants in Europe report significantly less of it (p. 26).
- Account takeover, 24 percent. A phished login opens the saved cards.
- Triangulation fraud, 15 percent. The buyer gets the goods, so the first complaint comes from a cardholder's bank.
- Payments with stolen card details. The survey has no row for them, so they carry no rank; UK Finance counts them as internet card fraud.
Phishing (37 percent) feeds types 4 and 6.
Why are online businesses such an easy target?
An online shop is vulnerable because its checkout never sees the buyer, and the card details typed into it may have been stolen months earlier from somebody else.
Europol's IOCTA 2026 calls fraud schemes the fastest growing area of organized crime. After the first fraud, stolen data is "often sold onwards" to other criminals (p. 17). Smaller dedicated platforms trade fraud tools such as one time password bypass kits and compromised card data (p. 12). Fraudsters use generative AI for call center scripts and to tailor social engineering, and Europol expects agentic AI to automate parts of the work (pp. 17 and 21).
UK Finance's Annual Fraud Report 2026 shows where the card data comes from. Data breaches, including hacks at third parties such as retailers, are a major driver of UK internet card fraud, and a card number taken in a breach stays usable "for months or even years" (p. 28). So the card at one shop's checkout can come from a breach at another.
6 common types of online business fraud: how do you stop each?
Table 1 gives each type its first red flag, the protection that stops it, what that control misses and who pays. The section numbers below are labels, not ranks.
Table 1. Six common types of online business fraud at a glance (as of November 2026)
| Type | When it shows | First signal in your data | Control that stops it | What the control misses | Who carries the loss | 2026 data point (source) |
|---|---|---|---|---|---|---|
| Card testing | Before the real fraud | Bursts of small authorizations and declines from one device, IP address or BIN range | Velocity limits per device, card and IP address; a bot check before authorization | Slow tests spread over days and devices | The tested merchant answers to its acquirer's enumeration threshold; the fraud loss lands where the card is used | 33% of merchants (MRC, Visa Acceptance Solutions, Verifi) |
| Friendly fraud | After delivery, at the bank | A disputed order sharing device, IP address and login with undisputed past orders | A descriptor customers recognize, delivery proof, evidence at the dispute | A first order with no history | The merchant, unless it wins the dispute | 36% met first party misuse (same survey) |
| Refund fraud and policy abuse | After delivery, at customer service | Repeat "not received" claims from one account, address or device | Refunds to the original payment method, claim limits, delivery photos, return inspection | Claims spread over many new accounts | The merchant; no card dispute to contest | 41% of merchants; 52% name false not received claims |
| Account takeover and new account fraud | At login or sign up | A new device, then a changed email or address, then a large order | Device and session scoring, step up when details change | A takeover run from the customer's own device | The merchant for disputed orders | 24% met account takeover |
| Interception and clean fraud with stolen cards | At checkout | A delivery change after dispatch; a first order from a device with no history | 3DS authentication, address checks, no carrier redirects on first orders | Clean fraud, where every detail matches | The merchant in a Visa card absent fraud dispute, unless 3DS authenticated the payment (most merchant categories) | £250.8m UK internet card fraud in 2025, up 9% (UK Finance) |
| Triangulation fraud | Weeks after delivery | A delivery name that differs from the cardholder's, across orders sharing a device | Linking orders by device, card and address; holding first orders where names differ | One clean looking order | The real shop that shipped the goods | 15% of merchants |
1. Card testing fraud
Card testing, also called enumeration, uses a bot to run stolen or generated card numbers through a low value checkout or donation form and learn which are live. The red flag is a burst of small authorizations and declines from one device, IP address or BIN range. Under Visa's VAMP fact sheet, acquirers must take steps to keep merchants below the program's enumeration thresholds: a ratio of 2,000 basis points, a fifth of all approved and declined authorizations, and 300,000 enumerated transactions in a month. A velocity limit per device, card and IP address, plus a bot check, is the control. A slow test spread over days gets past both. The fraud loss itself lands later, at whichever shop takes the live card.
2. Friendly fraud
Friendly fraud is a customer disputing a purchase they really made, which the survey calls first party misuse. It happens at the card issuer: the goods arrive, then the customer tells the issuer the payment was unauthorized or the parcel never came. Picture a subscriber (illustrative) who forgets an annual renewal, sees an unfamiliar descriptor and disputes the charge. The tell is a dispute on an order whose device, login and IP address all appeared on that customer's earlier, undisputed orders. A billing descriptor customers recognize, renewal notices, delivery proof and evidence filed at the dispute are the controls. A first order has nothing to match, and the merchant carries the loss unless it wins the dispute.
3. Refund fraud, policy abuse and item not received claims
Refund fraud is a false claim for money back: the parcel never arrived, the box came empty, or the return is used, damaged or a different item. It leads the 2026 list, and 61 percent of merchants saw it rise over the year. Asked which kinds they meet, most merchants (52 percent) name item not received claims, false claims that goods never arrived (p. 37). The older form is the overpayment, where a fraudster pays too much with a stolen card and asks for the difference back by bank transfer, so clean money leaves before the dispute arrives. Watch for repeat claims from one account, address or device. Refunds only to the original payment method, claim limits per customer, delivery photos and return inspection are the controls; claims spread over new accounts slip through. With no card dispute to contest, the merchant carries the whole loss.
4. Account takeover fraud and new account fraud
Account takeover is a criminal logging in as the customer, usually with phished credentials, then changing the email or address and spending saved cards or points. Europol calls the prominence of phishing and smishing "a direct adaptation" to the EU's shift to two factor authentication (IOCTA 2026, p. 21). New account fraud is the mirror case. Accounts are opened with stolen or invented details for sign up offers or stolen card purchases, the pattern behind e-commerce application fraud.
The warning sign is a new device, then changed contact details, then a large order. Scoring the whole session and stepping up when details change is the control. complead, formerly Sanction Scanner, runs onboarding and payment fraud checks on one engine, so a new account keeps the same risk score from its sign up form to its first order. A takeover from the customer's own device is the blind spot, and the merchant carries the disputed orders.
5. Interception and clean fraud with stolen cards
Older lists called this payment fraud and labeled it identity theft. Identity theft means taking over a person's identity, which this fraud never needs: a card number, expiry date and security code are enough. UK Finance put UK internet and e-commerce card fraud at £250.8 million in 2025, up 9 percent and the highest total since 2018 (p. 28). Interception fraud ships to the cardholder's own address so the address check matches, then redirects the parcel with the carrier. Clean fraud uses a stolen card with matching details, a fresh device and an ordinary basket, so mismatch rules let it through. The first sign is a delivery change after dispatch, or a first order from a device with no history. 3DS authentication, address checks and no carrier redirects on first orders are the controls. Visa's rules shield a 3DS authenticated payment from most card not present fraud disputes, so the exposure sits in payments sent without it.
6. Triangulation fraud
Triangulation fraud has three parties: a buyer, a scammer posing as a seller and a real shop. The scammer advertises goods cheaply, one of the "fake accommodation or merchandise listings" Europol sees in malicious advertising (IOCTA 2026, p. 21), takes the buyer's money, then orders the item from a real shop with a stolen card and the buyer's address. The buyer is satisfied, so nobody complains until the real cardholder's bank files a fraud dispute weeks later. It hit 15 percent of merchants, and MRC members report it significantly more often than other large merchants (p. 26). Look for a delivery name that differs from the cardholder's, across orders that share a device but not a card. Linking orders by device, card and address catches the pattern; one clean order does not show it. The real shop carries the loss.
Which two more types should online businesses watch?
Two more types come from partners rather than customers.
Affiliate fraud hit 22 percent of merchants (p. 25). An affiliate fakes clicks, sign ups or sales to collect commission, sometimes with stolen cards. Picture a coupon partner (illustrative) whose referred orders are refunded three times as often as the shop's average. The control is paying commission after the return window and tracking refunds and disputes per affiliate.
Merchant fraud and transaction laundering target marketplaces and payment providers. A seller takes payment and never ships, or runs another business's payments through its own account, which is transaction laundering. The control starts before a seller's first sale and keeps running after it. Hepsiburada, a leading ecommerce company in Türkiye, chose complead Fusion for merchant compliance: sanctions and PEP screening of each merchant at onboarding, daily rescreening and a merchant risk assessment with one risk score per merchant.
What recent cases show online business fraud at work?
In October 2025, Austrian, Estonian, Latvian and Finnish authorities, with Eurojust and Europol, arrested seven Latvian nationals who ran a SIM farm: at least 1,200 SIM box devices operating 40,000 SIM cards, with numbers registered to people from over 80 countries. More than 49 million online accounts were created through the service. Most fraud schemes that used it involved phishing, and the damage ran to millions of euros (IOCTA 2026, p. 19).
Each SIM card carried a phone number registered to a real person, so a sign up check that only confirms a texted code would pass every one. Shared devices, and accounts that act in step across sign ups, are what remains to catch.
Who pays for online business fraud, and what do the laws add?
Card scheme rules decide who carries a disputed card payment, and the law decides what the fraudster faces:
- Card rules. Under Visa's Core Rules of 24 October 2026, a card absent fraud dispute (condition 10.4) is invalid when the payment was authenticated through Visa Secure with EMV 3DS and sent with ECI 5 and the CAVV (p. 707), except in the high risk categories listed on p. 705. VAMP sets acquirers fraud, dispute and enumeration thresholds.
- US. Consumers report fraud to the FTC at ReportFraud.ftc.gov, and the FTC drew on those reports in its statement to Congress of 25 March 2026. The statement sets out no reporting duty for online shops.
- UK. In England, Wales and Northern Ireland, fraud by false representation is an offense under section 2 of the Fraud Act 2006, punishable by up to 10 years in prison on indictment (s. 1(3)). KYC and AML duties come from the Money Laundering Regulations 2017, which apply to the relevant persons listed in regulation 8, such as credit and financial institutions.
- EU. Under Article 97 of PSD2, payment service providers apply strong customer authentication when a payer starts an electronic payment, with elements that tie a remote payment to its amount and payee.
- Canada. Section 342(3) of the Criminal Code covers fraudulently using or trafficking in credit card data "whether or not the data is authentic", with up to ten years on indictment; section 342.1 covers unauthorized use of a computer, also up to 10 years (current to 19 October 2026).
Which online fraud numbers moved in 2026?
- March 2026. The merchant survey counted 3.8 fraud attack types per merchant on average, down from 4.2, and real time payment fraud fell from 45 to 38 percent of merchants (p. 25).
- 25 March 2026. In its Joint Economic Committee statement, the FTC said reported fraud losses had risen year over year for six years (p. 2).
- April 2026. The FBI's IC3 2025 report logged about 14 percent more non payment and non delivery complaints than in 2024, while the losses reported for them fell about 36 percent.
- June 2026. UK Finance reported UK internet card fraud up 9 percent, its highest total since 2018.
Older lists opened with a stolen card filed as identity theft. The 2026 survey opens with a refund claim, and two of the three most common types on this list need no stolen data at all. To find the type to fund first, count last quarter's refunds, disputes and declines by reason and by device; the largest pile points to the first control.
Checking the six types. Emirhan Akgün read the 2026 Global eCommerce Payments & Fraud Report, Europol's IOCTA 2026, UK Finance's Annual Fraud Report 2026, the FBI's IC3 2025 report, Visa's VAMP fact sheet and Core Rules, and the UK, EU and Canadian texts on 2 and 3 November 2026. Javid Khasayev (Senior Anti Fraud Specialist, GoldenPay) checked the controls and figures. Review date: 6 November 2026.