How we measure
CompliantHQ's scanner visits every site in a real browser and measures three areas: cookies and tracking before consent, accessibility according to WCAG 2.2 AA, and policy and terms documents against GDPR and Swedish law. Here is how it works.
Cookies and tracking
We open the site in an automated Chrome and record every cookie that is set and every request that goes out before the visitor has made a choice in the consent banner. We also try declining and see whether the tracking continues anyway. We compare this with the consent requirement in Chapter 9, Section 28 of the Swedish Electronic Communications Act and with GDPR. What we can tie to a known service for, say, statistics, advertising or embedded content counts as an issue. Cookies that we cannot tie to such a service with certainty, and that may be necessary for the site, are shown separately and are not counted.
Accessibility (WCAG 2.2 AA)
We run the tool axe-core from Deque Systems against every page. It is the same engine that Lighthouse uses. Where axe-core cannot determine whether a text is readable against its background, an AI model assesses a screenshot, and we only report the cases where the model is certain that the text is unreadable. In the study we show at most ten types of problems per site, the most serious first. If there are more, the number is marked with a plus sign. Requirements at the highest level, AAA, are not counted.
Policy & terms
We fetch the site's policy and terms pages (privacy policy, cookie policy, and terms of sale and use) and read them against GDPR and Swedish law. Most checks are rule-based, and some are made by an AI model. What we can establish from the text is reported as a remark. What we did not find, but which may be phrased in a way we did not catch, is reported separately as points to review and is not a claim that anything is missing. Which requirements are checked depends on what the site does, for example terms of sale only for e-commerce. In the study overview the area is therefore shown as a state, not as a number.
How often
The sites in a study are measured again at least every three months. The report page always shows the latest measurement, with the date at the top of the page. A site owner can ask for a new measurement sooner.
What we don't measure
We do not make legal assessments. That a site has zero issues in our measurement does not mean that it follows every rule, only that it passed our automated checks on the day we measured. The study compares a small number of pages per site, the five closest to the home page, so that large and small sites are measured alike. The design of the consent banner and what happens behind the site, such as routines and agreements, are not part of the study.
Questions or objections
We report what the tools measured and judge no one. If you think a measurement is wrong we want to know, so that we can correct it. Email hello@complianthq.ai.