Chargeback statistics for 2026, with a named source for every number
If you need one chargeback statistic to quote, use this: global payment card fraud losses reached $33.41 billion in 2024 (Nilson Report, 2024 data, checked 2026-08-10).
If you need a second, roughly 20% of all fraudulent disputes are first-party misuse, the behaviour most people still call friendly fraud, climbing to 30% among high-volume online merchants (Visa, citing the 2025 Global eCommerce Payments & Fraud Report, checked 2026-08-10).
This page collects the chargeback statistics worth quoting for 2026, each traced to a named source with a year. That constraint is the point: a figure you cannot trace is a figure you cannot defend when someone asks. If the mechanic itself is new to you, start with what a chargeback is, then come back for the data.
The numbers at a glance
| Statistic | Figure | Source (checked 2026-08-10) |
|---|---|---|
| Global card fraud losses, 2024 | $33.41 billion, down 1.2% | Nilson Report, 2024 data |
| US share of fraud vs volume | 41.87% of losses on 26.31% of volume | Nilson Report, 2024 data |
| Cost per $1 lost to fraud, US merchants | $4.61 | LexisNexis True Cost of Fraud, 15th ed., Apr 2025 |
| Friendly fraud share of fraudulent disputes | ~20% globally, up to 30% high-volume | Visa, 2025 Global eCommerce Payments & Fraud Report |
| Merchants reporting rising first-party misuse | 64% | MRC 2026 Global eCommerce Payments & Fraud Report |
| Friendly fraud as a share of chargebacks | 22% in 2026, 28% by 2031 | Juniper Research, Jun 2026 |
| Friendly fraud transaction value | $8.1B in 2026 to ~$16B by 2031 | Juniper Research, Jun 2026 |
| Average representment win rate | 44.6% of disputes fought | Chargebacks911, 2026 Chargeback Field Report |
| Net recovery rate, all chargebacks | 10.7% | Chargebacks911, 2026 Chargeback Field Report |
| VAMP Excessive Merchant threshold | 150 bps from 1 Apr 2026, US/CA/EU/AP | Visa Acquirer Monitoring Program fact sheet |
How much do chargebacks and fraud cost?
Global card fraud losses were $33.41 billion in 2024, a dip of 1.2% on the prior year, set against total card volume of $51.920 trillion (Nilson Report, 2024 data, checked 2026-08-10). It measures gross fraud losses across issuers and acquirers, not chargeback volume specifically.
The United States carries a lopsided share. It accounted for 26.31% of worldwide card volume in 2024 but 41.87% of global fraud losses (Nilson Report, 2024 data, checked 2026-08-10). If you sell into the US, the average understates your exposure. Nilson projects losses reaching $41.06 billion by 2030.
Fraud costs a merchant more than the fraud itself. For every $1 lost to fraud, US ecommerce and retail businesses incur $4.61 in total costs once fees, labour, and lost goods are counted, with Canada at $4.52 (LexisNexis True Cost of Fraud Study, 15th edition, published 2 April 2025, checked 2026-08-10). That multiplier is why a disputed order is never just the order value.
How common is friendly fraud?
This is the most misquoted area in the category, so the source matters more than usual.
Visa puts first-party misuse at around 20% of all fraudulent disputes globally, and up to 30% for high-volume online merchants (Visa, citing the 2025 Global eCommerce Payments & Fraud Report, checked 2026-08-10). Those are the most conservative, best-attributed numbers I found. Much higher figures circulate; the ones I chased traced back to surveys with no published methodology.
The trend is up, by merchant self-report. 64% of merchants report increasing rates of first-party misuse, and a quarter of them report increases of 25% or more (MRC 2026 Global eCommerce Payments & Fraud Report, surveying 1,278 merchant professionals across 37 countries, published 18 March 2026, checked 2026-08-10).
Forecasters expect the share of chargebacks to keep rising. Friendly fraud is projected to move from 22% of chargebacks in 2026 to 28% by 2031, with the transaction value growing from $8.1 billion to nearly $16 billion over the same span, a 96% increase (Juniper Research, published June 2026, checked 2026-08-10).
A large slice of this is avoidable confusion rather than deliberate abuse: a customer who does not recognise a billing descriptor, or a household member's purchase. That distinction is the subject of first-party fraud, and closing the confusion gap is what preventing avoidable disputes is aimed at, which is a different job from fraud screening.
What is the average chargeback win rate?
Two numbers matter here, and quoting only the first is how merchants mislead themselves.
Merchants win about 44.6% of the chargebacks they represent (Chargebacks911, 2026 Chargeback Field Report, checked 2026-08-10). That is the figure most people mean by win rate, and it counts only disputes actually fought.
Net recovery tells the fuller story at 10.7% (Chargebacks911, 2026 Chargeback Field Report, checked 2026-08-10). Net recovery is recovered dollars over all chargebacks issued, so it absorbs the disputes nobody fought and the first-cycle wins lost to second-cycle escalation. These are self-reported survey figures, so treat the gap between them as directional.
The low net-recovery number is mostly an operations problem, not a merits problem: cases go unfiled because nobody owns the deadline. We built Reclaim to file representments at no install cost and no cut of the recovery, precisely so the winnable cases stop expiring in a queue.
What are the 2026 VAMP thresholds?
Visa consolidated its fraud and dispute monitoring into a single Visa Acquirer Monitoring Program, and the current fact sheet (checked 2026-08-10) is the number to trust over any blog, including this one.
The VAMP ratio counts fraud plus disputes over settled transactions. Formally, it is the count of Fraud (TC40) plus Disputes (TC15) on card-not-present transactions, divided by the count of settled transactions (TC05), effective 1 June 2025. Disputes resolved through pre-dispute solutions and fraud qualifying for Compelling Evidence 3.0 are excluded from the count.
The Excessive Merchant threshold falls to 150 basis points on 1 April 2026 for the US, Canada, EU, and Asia Pacific, down from 220 basis points, with LAC already at 150 and CEMEA at 220 (Visa Acquirer Monitoring Program fact sheet, checked 2026-08-10). A merchant enters the program only above a minimum monthly volume of 1,500 fraud-plus-dispute events in those regions. At the acquirer-portfolio level, Above Standard is 50 basis points and Excessive is 70.
Why these numbers vary so much
Chargeback statistics scatter wildly across sources, and the reason is definitional, not sloppy.
Each source measures a different population. Nilson counts gross card fraud losses across the whole system, LexisNexis surveys risk executives on total cost of fraud, Juniper forecasts from a model, and Chargebacks911 aggregates a merchant survey. So a $33 billion fraud figure and an $8 billion friendly-fraud figure are not in tension. They count different boxes.
Friendly fraud shares diverge because the denominator moves. Twenty percent of fraud-coded disputes is a different quantity from twenty percent of all disputes, which is different again from twenty percent of chargeback value. A number without its denominator is decoration.
Win rates are self-reported. The 44.6% figure comes from merchants describing their own results, and the population that answers a chargeback survey is not a random sample of all merchants. Treat any single win rate as a starting hypothesis.
The practical rule: distrust any chargeback statistic presented without a named source and a year. If a roundup cannot tell you who counted the number and when, it is quoting another roundup, and the trail usually dead-ends at nobody.
Frequently asked questions
How much do chargebacks cost merchants in 2026?
There is no single chargeback cost figure, so use the traceable ones. Global card fraud losses were $33.41 billion in 2024 per the Nilson Report, and US ecommerce and retail merchants incur $4.61 in total cost for every $1 lost to fraud per the LexisNexis True Cost of Fraud Study, 15th edition, April 2025. On a single order, expect the order value at risk plus a dispute fee, commonly $15 to $25, plus 20 to 45 minutes of evidence work.
What percentage of chargebacks are friendly fraud?
Visa estimates first-party misuse at around 20% of all fraudulent disputes globally, up to 30% for high-volume online merchants, citing the 2025 Global eCommerce Payments & Fraud Report. Juniper Research projects friendly fraud rising from 22% of chargebacks in 2026 to 28% by 2031. Higher percentages circulate but rarely name a methodology.
What is the average chargeback win rate?
Merchants report winning about 44.6% of the chargebacks they represent, but the net recovery rate across all chargebacks is just 10.7%, both from the Chargebacks911 2026 Chargeback Field Report. The gap exists because many disputes go unfought and some first-cycle wins are reversed at second cycle. These are self-reported survey figures, so treat them as directional.
What is the 2026 VAMP threshold?
Visa's VAMP Excessive Merchant threshold drops to 150 basis points, or 1.5%, on 1 April 2026 for the US, Canada, EU, and Asia Pacific, down from 220 basis points, per Visa's Acquirer Monitoring Program fact sheet. LAC is already 150 and CEMEA is 220. The ratio counts fraud plus disputes over settled transactions, gated by a minimum of 1,500 events a month in those regions.
What to do with these numbers
Benchmarks are for orientation, not decisions. The two numbers that should drive yours are your own, and you already hold both.
Pull your dispute rate, counted the way VAMP counts it: fraud plus disputes over settled transactions, measured against the 150 basis point line for your region. Then pull your net recovery rate, recovered dollars over all chargebacks issued, not wins over cases fought, which flatters you the moment you start declining the hard ones.
Compare those two to the figures above. If your dispute rate sits near the threshold, prevention and deflection are worth more than any recovery. If your net recovery is well under 10%, the problem is usually unfiled cases, not unwinnable ones. Your own two numbers beat every benchmark on this page.