An accessibility overlay is a third-party script you paste into your site, sold on the promise that it repairs your accessibility failures at runtime and covers your legal obligation. The obligation is real: the European Accessibility Act (Directive (EU) 2019/882) has applied since 28 June 2025, and Turkey's Circular 2025/10, published in the Official Gazette on 21 June 2025, gives e-commerce providers under Law No. 6563 two years to reach WCAG 2.2 Level A. The coverage promise is what does not hold. In UsableNet's 2026 midyear data, roughly 20% of companies sued over digital accessibility already had a widget or overlay installed. This guide gives you eight criteria for choosing between an overlay and code-level remediation, five questions to score any vendor pitch, and the sequence we run instead.
What an overlay is, and which half makes the compliance claim
Every overlay bundles two different things behind one script tag, and they deserve separate judgement.
- A user-preference toolbar: larger text, higher contrast, wider spacing, a reading mask. Harmless and mostly redundant, since browsers and operating systems already ship these controls.
- Automated DOM repair: a script that scans the rendered page and injects guessed alternative text, ARIA roles, labels and focus handling. This is the part sold as conformance, and the part that fails.
On 3 January 2025 the US Federal Trade Commission announced an order requiring accessiBe to pay $1 million for claiming its accessWidget product "could make any website compliant with the Web Content Accessibility Guidelines". The FTC found those claims "false, misleading, or unsubstantiated", and found the company had formatted paid articles and reviews to look like independent opinion. The order bars it from claiming the product can make any website WCAG-compliant without evidence. Nothing equivalent has happened in the EU or Turkey, but the finding goes to the category's central claim, not to US law.
The question a regulator asks is what your conformance evidence is
Under the EAA, presumption of conformity comes from harmonised standards cited in the Official Journal. Today that is still EN 301 549 v3.2.1 (2021), mapping to WCAG 2.1 Level AA. Version 4.1.1, published in September 2026, adopts WCAG 2.2 and drops the obsolete 4.1.1 Parsing criterion, but carries no legal weight until the Commission cites it. Budget against v3.2.1 and design so the WCAG 2.2 additions are already satisfied.
Germany has already answered the overlay question inside that framework. The BIK BITV-Test methodology, used to award the German accessibility seal and aligned to EN 301 549, states that because of the interactions such a tool introduces, "Konformität für eine Seite mit Overlay-Tool, auch wenn es im Default-Zustand bleibt, nicht sichergestellt werden" kann: conformity cannot be assured for a page using an overlay tool, even in its default state. Sites running one cannot publish results or carry the seal.
Turkey works the same way. Circular 2025/10 is monitored against a Level A checklist built from WCAG 2.2, reported by the Ministry of Family and Social Services as 31 principles and 122 questions. Compliant organisations receive the Erişilebilirlik Logosu, valid two years; non-compliant sites have that status made public.
The common thread: the page is assessed as delivered to the user. An overlay is not outside that assessment. It is one more script inside it.
Eight criteria for the decision
| Criterion | Overlay alone | Code-level remediation |
|---|---|---|
| 1. Accepted as conformance evidence | No. BIK in Germany refuses to certify pages running one. | Yes. This is what EN 301 549 and the Turkish checklist assess. |
| 2. Keyboard and screen reader path through checkout | Unreliable. Focus order, error association and custom controls are what automated repair cannot infer. | Fixable, and human-testable in one session per flow. |
| 3. Single-page apps and dynamic content | Weak. The script repairs a snapshot; React, Vue and Angular re-render past it. | Handled at component level, so it survives re-renders. |
| 4. PDF invoices, canvas, SVG, video | Out of scope entirely. | In scope, and usually the largest hidden cost. |
| 5. Native mobile apps | Not applicable. Both regimes cover apps; a web script cannot reach them. | Applies, via the platform accessibility APIs. |
| 6. Time to first improvement | Hours. The genuine advantage. | Days to weeks for the first tranche. |
| 7. Cost shape | Recurring licence, indefinitely, no asset created. | One-off work plus retest, then a regression gate. |
| 8. Residual risk | A regulatory finding against the core claim, litigation exposure, public listing in Turkey. | Ordinary delivery risk: work slips, or a release regresses. |
Only row 6 favours the overlay, for a reason unrelated to conformance. If you want something visible while the real work runs, say that out loud rather than calling it compliance.
Five questions to score any vendor pitch
Score each 0, 1 or 2. Under 7 out of 10, buy it as a convenience feature or not at all, but not as a compliance measure.
- Will you put in the contract that your product alone brings this site into conformance with EN 301 549, and indemnify us against a contrary finding? 2 = yes in writing. 1 = yes with named exclusions. 0 = the claim lives only in the sales deck.
- Name a recognised testing body that certified a comparable site while your script was installed. 2 = a named body and a reachable reference. 1 = an internal report. 0 = nothing.
- List the success criteria your product remediates, and the ones it does not. 2 = both lists, criterion by criterion. 1 = partial. 0 = "we cover WCAG".
- Does the product detect assistive technology use, what is stored, and where? 2 = no detection, or detection with consent and EU/Turkey data residency. 1 = disclosed but stored elsewhere. 0 = unclear. Inferring disability status is special-category data under GDPR and KVKK.
- If we remove the script tomorrow, does the site return to its prior state? 2 = yes, plainly. 1 = yes with caveats. 0 = evasion, meaning you bought a dependency rather than a fix.
The sequence we run instead
- Automated baseline per template, not per page. One scan each for home, category, product detail, cart, checkout, account and static content. The WebAIM Million analysis of one million home pages in February 2026 found 95.9% with detected WCAG 2 failures and 56.1 errors per page on average, in six patterns: low contrast (83.9%), missing image alternative text (53.1%), missing form labels (51%), empty links (46.3%), empty buttons (30.6%) and missing document language (13.5%). Five of the six are design-system fixes that ship once and cover thousands of URLs.
- Manual pass on the revenue path only. Keyboard-only and screen reader traversal of search, product detail, add to cart, checkout and account creation. This is where the failures automation cannot see live, and where one costs an order.
- Fix in code, ordered by whether the failure blocks completion. A blocked checkout step outranks twelve contrast warnings in a footer.
- Retest, then write the statement. An EAA accessibility statement for the EU; the Level A checklist answers that support a logo application in Turkey.
- Put a regression gate in the pipeline, or the audit you paid for decays within two quarters.
[INTERNAL DATA NEEDED: Switas audit sample, share of overlay-equipped Turkish e-commerce sites still failing keyboard-only checkout completion, with sample size and period.]
None of this makes the toolbar half illegitimate. A clearly labelled preference panel on a site that already conforms is a reasonable product decision. Budget it as a convenience feature, and keep it out of the conformance conversation.
Where this guide breaks down
- There is no controlled public benchmark of overlay effectiveness. Percentages circulating on both sides are vendor or advocacy figures, not results from a published, replicable test. The strongest independent work is qualitative or survey-based: Daniela Kubesch's 2024 master's thesis (three overlays, 21 participants with visual impairments, concluding they cannot meet WCAG 2.1 AA); the Overlay Fact Sheet, signed by more than 1,000 practitioners; and WebAIM's 2021 survey, in which 67% of 758 practitioners rated overlays not at all or not very effective. Treat any precise "fixes X% of issues" figure, from any direction, as unverified.
- The German position is one testing body, not EU law. BIK's refusal shows how auditors think; it is not a Commission ruling.
- Circular 2025/10 does not mention overlays. How Turkey's monitoring commission treats a site running one is untested; we reason from the checklist's structure, not a published decision.
- This is not legal advice. Disproportionate burden and microenterprise exemptions exist under the EAA (a microenterprise being fewer than 10 persons with turnover or balance sheet total not exceeding EUR 2 million). Whether they apply is a question for counsel.
FAQ
Marked for FAQPage schema.
Does installing an accessibility overlay make my site WCAG compliant?
No. The FTC ordered accessiBe to pay $1 million in 2025 over exactly that claim. Conformance is assessed on the page as delivered, and an overlay is part of that page.
Which standard do I have to meet for the European Accessibility Act?
EN 301 549 v3.2.1, which maps to WCAG 2.1 Level AA, because that is the version cited in the Official Journal. Version 4.1.1, published in September 2026, moves to WCAG 2.2 but carries no legal weight until it is cited.
Turkey's circular says Level A. Is that enough?
It is what Circular 2025/10 monitors against. If you also sell into the EU you are in EAA scope, where the reference is Level AA. Build to the higher of the two once rather than remediating twice.
We are outside the EU. Does the EAA reach us?
The obligation follows the market, not the head office. Services provided to consumers in the EU are in scope wherever the provider is established, which is why Turkish exporters face two deadlines at once.
Can an overlay reduce our litigation risk?
The data does not support that. In UsableNet's 2026 midyear figures about 20% of companies sued had a widget or overlay installed, and the same report projects roughly 6,000 US accessibility lawsuits for 2026.
Is there a privacy problem with these tools?
Potentially. Products that detect assistive technology use are inferring disability status, which is special-category data under GDPR and Turkey's KVKK. Ask where that signal is stored, for how long, and on what legal basis.
Run the scan before you take the meeting
Run your home page, one product detail page and your checkout through our free scanner at wcag.switas.com, at the version and level your obligation requires, and see how much of the gap is template-level. If it is bigger than a script tag, talk to us about a full audit against EN 301 549 and the Turkish Level A checklist. What you get back is a remediation order, not a licence.
Sources
- Directive (EU) 2019/882 (European Accessibility Act), EUR-Lex
- EN 301 549 updated, AccessibleEU, September 2026
- Accessibility standard, latest changes, European Commission
- BIK BITV-Test, Beschreibung des Prüfverfahrens
- FTC order against accessiBe, 3 January 2025
- Circular 2025/10, Official Gazette 21 June 2025, No. 32933
- Ministry of Family and Social Services on Circular 2025/10
- The WebAIM Million, February 2026
- WebAIM Practitioner Survey #3, March 2021
- Overlay Fact Sheet
- Kubesch, The Impact of Web Accessibility Overlays, 2024
- UsableNet, Digital Accessibility Lawsuits in 2026
