Accessibility statements under ZDSMA: required content and the limits of self-assessment

ZDSMA binds state bodies, bodies of self-governing local communities, and bodies governed by public law under the public procurement act. In practice: ministries and agencies, all 212 municipalities, public institutes, schools, kindergartens, health centres and libraries. The statement must be published in an accessible form on the website itself.

Websites published after 23 September 2018 had to comply by 23 September 2019. Existing websites had until 23 September 2020, and mobile applications until 23 June 2021. There is no remaining transition period.

The statement must give an actual assessment of the site's compliance — compliant, partially compliant or non-compliant. It must identify inaccessible content with the reason, whether disproportionate burden or content outside scope, and where possible an accessible alternative.

It must also record the date of preparation or last review, the assessment method used, and contact details with the feedback and enforcement procedure.

The assessment method must be stated. Two are used in practice: self-assessment by the obligated body, and external review by a third party. The law does not require external review — self-assessment is a permitted method and by far the most common.

The difference is not legal permissibility but evidence. A self-assessment without recorded measurement is an assertion. When a user complaint or a supervisory question arrives, the gap between an assertion and a documented measurement with evidence is substantial.

Most statements say “partially compliant”, which is often an honest description. The problem is that the statement rarely says which content is non-compliant, why, and when it was last checked. A general claim of disproportionate burden without an assessment of that burden is the weakest part of a typical statement.

Stale dates are the second problem. A statement prepared in 2020 and untouched since describes a site that has changed repeatedly. The review date is one of the few facts anyone can verify in seconds.

A large share of Slovenian municipal websites run on shared platforms or shared templates. That has two consequences statements usually miss.

First, a defect in the shared template — navigation contrast, or focus behaviour in a menu — repeats across every site using it. A single body's self-assessment is then limited: the finding is real, but the fix is not in its hands. Second, that is exactly why fixing it at the vendor is the most efficient path — one change removes the same barrier for every site on the platform.

It is worth separating, in the statement, the content the body controls from the framework the vendor provides, and passing concrete evidenced findings to that vendor.

Measure a representative set of pages — home, search, a listing, a single item, a form, and any page where a user submits something. Record what was measured, what could not be verified, and against which version and date.

Then rewrite the statement so it names concrete inaccessible content instead of a general formula, states the method and date, and separates own content from the platform. The sample report shows the shape of evidence you can cite; scope and cadence are described under pricing.