Two organisations count US website accessibility lawsuits every year, and their 2026 mid-year numbers are thousands apart. Seyfarth Shaw reports 5,006 federal ADA Title III suits in the first half of the year. UsableNet projects roughly 6,176 digital accessibility suits for the full year. Both are right, because they count different things, and the gap between them is the most useful part of the story. This article explains what each tracker measures, who is actually being sued, what changed in the law this year, and the thirty-minute check that tells a small business where it stands.
This article is educational content, not legal advice. If you have received a demand letter or a complaint, talk to a lawyer before you respond. Lawsuit counts below come from the published reports of Seyfarth Shaw (adatitleiii.com) and UsableNet, and each figure is attributed to its source.
The two numbers, and why they differ
Seyfarth Shaw publishes the longest-running count. Its mid-year report, dated 9 September 2026, puts first-half 2026 filings at 5,006 federal ADA Title III lawsuits, compared with 4,575 in the first half of 2025, a 9% increase. The firm notes the first-half average for the previous four years was 4,463, so 2026 is running above trend. Two scope details matter: Seyfarth counts federal court only, and the Title III total includes physical-access cases (ramps, parking, counters), not just websites. Its website-specific count is published separately: 3,117 federal website accessibility suits in 2025, up from 2,452 in 2024.
UsableNet counts federal and state courts, and only digital cases (websites, apps, video). Its 2026 mid-year report projects approximately 6,176 lawsuits for the year, based on filings recorded through 14 June, and describes a continuing shift toward state courts, New York in particular.
So the numbers are not in conflict. One is a federal, all-access count; the other is an all-court, digital-only count. If you want the trend, read Seyfarth. If you want the exposure profile of an online business, read UsableNet. Any article that quotes one number as "the" number without saying which scope it uses is giving you half the picture.
Who actually gets sued
The most persistent myth in this area is that plaintiffs only go after big brands. UsableNet's defendant-revenue breakdown for 2026 says otherwise:
- 68% of defendants had annual revenue under $50 million
- 22% had revenue between $50 million and $1 billion
- 10% had revenue over $1 billion
The same report notes that the share of defendants with revenue above $25 million has been rising, from 27% in 2023 to 33% in 2024 and 36% in 2026. Read those two facts together: larger companies are being sued more often than they used to be, and smaller companies still make up two thirds of the defendants.
Sector matters more than size. UsableNet reports that 79% of 2026 filings target e-commerce. The reason is practical. A claim needs a barrier to a good or service, and a checkout that cannot be completed with a screen reader is the clearest barrier there is. A brochure site for a plumbing company is a weaker case than a store with a cart.
Geography is concentrated too. In Seyfarth's first-half 2026 federal count, California led with 2,426 Title III filings, followed by Florida (709), Illinois (505), New York (485) and Minnesota (124). For website-only federal cases in 2025, the order was different: New York 1,021, Florida 961, Illinois 585. The difference is the physical-access caseload in California and the state-court preference of New York website plaintiffs. Either way, if you sell to customers in New York, Florida or Illinois, you are in the jurisdictions where most website claims are filed, regardless of where your company is registered.
One more structural fact from the Seyfarth data: filings are concentrated among a small number of plaintiff firms. Of the 505 Illinois filings in the first half of 2026, 336 came from a single firm, and 121 of Minnesota's 124 came from another. That is not a reason to dismiss the cases. It is a reason to understand that these firms run a repeatable process, and that the process starts with an automated look at your site.
How a case typically starts
For a small business, the first contact is rarely a filed complaint. It is usually a demand letter: a letter from a law firm stating that a named client with a disability tried to use your site, listing the barriers encountered (often phrased in WCAG terms: missing alternative text, unlabeled form fields, keyboard traps, missing focus indicators), and proposing a settlement that includes remediation and attorney's fees.
Three things are true about these letters that are worth knowing before one arrives:
- The barriers listed are usually real. They are found with the same automated tools anyone can run, plus a short manual pass. Whether they amount to a legal violation is a separate question, but "our site is fine" is rarely a complete answer.
- The letter is not the lawsuit. Many are resolved without a filing. How you respond, and how quickly you can show documented remediation, affects the outcome. That is a conversation for your lawyer, not for a blog post.
- The cheapest moment to fix the site is before the letter. Every item on the list is a known, fixable defect. The cost of fixing them is the same before and after; the legal cost is not.
What changed in the law in 2026
The DOJ Title II extension (government websites)
In April 2026 the Department of Justice extended the compliance dates for its 2024 rule requiring state and local government websites and apps to meet WCAG 2.1 AA. The new dates are 26 April 2027 for larger public entities (population of 50,000 or more) and 26 April 2028 for smaller ones and special districts. The National Federation of the Blind has challenged the extension in court; as of this writing we have not found a ruling.
If you are a private business, this extension does not touch you. Title II covers governments. Private website claims are brought under Title III and state law, and the first-half 2026 numbers above show that caseload growing, not pausing. If you build websites for cities, counties or school districts, the clock moved, but it is still a clock.
Missouri's notice-and-cure law
Missouri enacted a law in May 2026, operative from 28 August 2026, aimed at what the legislature called abusive website litigation. It creates a rebuttable presumption that a suit is abusive where the defendant received written notice of the alleged website violation and initiated substantial steps to correct it within ninety days, and it lets Missouri-resident defendants (and the state Attorney General) act against plaintiffs and their lawyers. It is a state law and does not change the federal ADA. Its practical lesson, though, applies everywhere: a documented, dated remediation record after notice is the thing every one of these frameworks rewards.
Overlays: the regulatory record
In April 2025 the Federal Trade Commission finalised a consent order against the largest accessibility-overlay vendor, with $1 million in monetary relief, over claims that its widget could make any website WCAG-compliant. That order is the clearest statement a US regulator has made on the subject: a script that paints controls over a site without fixing the underlying code is not a compliance method. If a vendor's pitch is "install this and you're covered," the regulator has already said it isn't.
What an automated scan can and cannot tell you
Every demand letter starts with an automated scan, so it makes sense to run one first. It also makes sense to know its limits, because the limits are where most of the remaining risk lives.
Automated tools, including ours, reliably find a portion of WCAG failure types: missing or empty alternative text, insufficient colour contrast, form fields without labels, buttons without accessible names, broken heading structure, missing page language, duplicate IDs. Estimates of the share of issues a machine can catch vary with the ruleset and the page; figures from about a quarter to around 40% are commonly cited, and none of them reach half.
What a scanner cannot judge: whether alternative text actually describes the product, whether focus order makes sense, whether a custom dropdown works with arrow keys, whether an error message is announced to a screen reader, whether a cookie banner traps keyboard focus, whether checkout can be completed at all without a mouse. Those are exactly the failures that appear in demand letters, because the plaintiff's expert did them by hand.
A clean automated report is therefore a floor. It means the obvious defects are gone. It does not mean the site is accessible, and any tool that says "compliant" on the strength of a scan is overclaiming.
The thirty-minute check
Here is the sequence we recommend to any small business, in the order that finds the most risk per minute. None of it requires a purchase.
- Unplug the mouse (10 minutes). Open your site, press Tab, and keep pressing. Can you see where you are at every step? Can you open the menu, reach a product, add it to the cart, and get to the payment form using only Tab, Shift+Tab, Enter, Space and the arrow keys? Does a cookie banner or chat widget ever swallow the focus? If you get stuck, a screen-reader user gets stuck in the same place, and that place is what the letter will describe.
- Run a real scan on three pages (5 minutes). Home, one product or service page, and checkout or your main form. Any axe-based tool will do: Lighthouse in Chrome, axe DevTools, or TrueConform's free scan. Export the list.
- Sort by what blocks a purchase (5 minutes). Unlabeled inputs and unnamed buttons in checkout come first. Then keyboard traps and invisible focus. Then images of products with missing or useless alt text. Then contrast. Decorative flourishes last.
- Check the third-party pieces (5 minutes). Chat widgets, review carousels, cookie banners, embedded video players and app-store plugins are the most common source of failures on otherwise well-built stores, because nobody on your side wrote them. Tab into each one. If it fails, ask the vendor for their accessibility conformance report, and if they don't have one, that is an answer too.
- Write it down (5 minutes). Date the scan. Date each fix. Publish a short, honest accessibility statement that says what you tested, what you fixed, what is still open, and how to reach a human if something doesn't work. A dated record is what every notice-and-cure framework, and every settlement negotiation, asks for first.
If you sell into the EU as well
The US picture is private litigation. The EU picture is regulators and courts, and it is moving in the same direction. A French court ruled in June 2026 that a retailer's 71% conformity did not satisfy the European Accessibility Act, treating accessibility as an obligation of result; the reasoning is in our write-up of the Carrefour ruling. Sweden's regulator, PTS, measured 576 websites across nine sectors in 2025 and found deficiencies in every sector, with up to 86% of public-sector sites showing issues that can exclude users; its report notes that automated measurement understates the problem, for the same reasons set out above. If your store serves EU customers, the keyboard test and the scan are the same; the enforcement path is different.
And if your site runs an AI chat widget, there is a separate EU duty to tell visitors they are talking to an AI, in force since 2 August 2026. That one is covered in our Article 50 guide for small businesses.
What the numbers do not say
A few cautions, because this topic attracts inflated claims. The lawsuit counts above are filings, not findings of liability; most cases settle. The trackers count differently, so comparisons across sources are unreliable unless the scope is stated. "Risk" figures attached to any single site are marketing, not actuarial data; nobody can tell you your probability of being sued. And a scan score, ours included, is not a compliance certificate.
What the numbers do support is narrower and more useful. Filings are rising. Most defendants are small. Most cases involve a checkout. The defects named in demand letters are findable in advance with tools you already have, and the manual check that catches the rest takes less time than reading this article did. The businesses that come out of 2026 without a letter will mostly be the ones that did the thirty minutes.