What the NeuroDesignTest measures — and what it deliberately does not
Ten areas, 53 checks, a report in your inbox. Nine checks are measured, the rest rely on judgement — and we say so plainly. How the test works and how to read its score.
You cannot see your own website. That is not a criticism, it is how brains work: you know where everything is, you know every sentence, and you know what you meant. A visitor knows none of it. They see a surface, decide within fractions of a second whether it concerns them, and only then read — if at all.
That is why opinions help so little here. „I like it" and „it brings enquiries" are two different questions, and the first does not answer the second. The NeuroDesignTest is our attempt to answer the second one in a way you can actually act on.
What the test does
You give us an address and your email. After you confirm the link in the email, our software opens the page, looks at it the way a visitor would, and then works through a fixed grid: ten areas with 53 checks in total. Every check gets points, and every check gets a reason. At the end there is a number from 0 to 100 and, more importantly, a list of what is missing.
The ten areas follow the order in which a decision forms — from the first screen through trust and proof to the call to action. It is the same grid we use on client projects; we did not invent it for the test, we built the test out of it.
Nine checks are measured. The rest rely on judgement.
This is the part no provider likes to write, and the only one that makes the report usable.
Nine checks are measured and recalculated: colour contrasts to WCAG, positions in the page layout, page weight, meta tags. Those values are identical run after run. If something changes there, something changed on your site — not in our assessment.
The rest rely on an assessment. Whether a headline really breaks a pattern, whether a proof reads as proof, whether the next step is unambiguous — none of that can be measured with a ruler. A language model rates those points against our brief. Two runs on the same page typically differ by several points.
From that follows the only rule you need when reading it: only a difference of about ten points means anything has really changed. Going from 61 to 64 after a rebuild proves nothing. Going from 61 to 74 does.
How to read the score
The total is the least interesting line in the whole report. It works as a starting value and not much else. What tells you something is the level below it:
- Which areas are red. Ten areas, three colours. Two red areas tell you more about your next fortnight than any total.
- Which individual checks are missing. Each carries a reason in the same language you use with customers — not „CTA hierarchy suboptimal", but which sentence is missing and why.
- What changed between two runs. The comparison per check tells you something even when the total stands still.
And the three most important steps are at the end of the report, ordered by what pays off first.
What the test is not
It is not a warranty that your site has any particular property. It is not an expert opinion. It says nothing about the legal compliance, accessibility or security of your site — even though it measures contrasts, it does not replace an accessibility audit. And no commercial result can be derived from the score: nobody can honestly tell you that 80 points means twice as many enquiries as 40.
What it can do: show you where an outside eye gets stuck, with the reasoning attached.
Where it runs and what happens to your data
The evaluation runs on our own hardware in Austria. No data goes to third-party AI services. We delete the screenshots of your site right after the evaluation.
You get the report not as an attachment but as a page with a link only you have — plus a PDF to download. After 14 days we delete the file and the page. An attachment sitting in other people-s inboxes could never be withdrawn; a promise to delete it would then be a lie.
The test only runs on explicit request, and only after you confirm the link in the email. Anyone requesting an analysis of a website thereby confirms that they are entitled to do so.
What to do with the result
Preferably nothing grand. Take the two reddest areas and the first of the three steps, change exactly that, and run the test again in two weeks. If the affected check moved, it was the right change. If it did not, in two weeks you have a finding instead of a hunch.
The test is free, and you usually get the result in under 30 minutes. Whatever you take away is yours — even if we never work together.
Request the website test — or, if you would rather go through the report together: 30 minutes by video.
Structure and limits of the grid: as of September 2026. The statements on measurement, variance and deletion periods appear verbatim in the legal notice.
Source: Luminant Mind e.U., Wien — Aufbau und Grenzen des eigenen Prüfrasters, Stand September 2026
Want to know where your own site stands?
We look at your site and send you an assessment with a score — 10 areas, with a reason given for each point.
- our own software, not a third-party tool
- evaluated locally in Austria
- analysis without third-party AI
- revocable at any time
Keep reading

Your Website Looks Professional. So Why Is No One Contacting You?
A good first impression keeps visitors on the page. It does not answer the questions that decide whether they get in touch.

Why your website doesn't bring clients — and what brain research says about it
95% of all buying decisions are made unconsciously. If your web design ignores this principle, you systematically lose clients.

The 50-Millisecond Test: How first impressions decide trust
Researchers at Carleton University have shown: in 50ms your website is judged. What needs to happen in these milliseconds.