We scored our own site — then fixed it
The Answer Engine makes your site the one ChatGPT, Gemini and Google's AI Overviews recommend. Before we sold it, we ran it on ourselves. Here's the whole scorecard — every signal, before and after, including the ones we haven't finished.
out of 100 — AI-search readiness
Ten signals, one hundred points
We fetch a site the way an AI crawler does — raw HTML, no JavaScript — and score ten signals that decide whether a model can find, read, trust and quote it. Here's how our own site moved.
| Signal | What it measures | Max | Before | After |
|---|---|---|---|---|
| Structured data | Schema.org JSON-LD — Organization, Service, FAQ, Person — present, valid and cross-linked | 15 | 0 | 14 |
| AI-crawler access | robots.txt welcomes GPTBot, ClaudeBot, PerplexityBot and Google-Extended; sitemap present | 12 | 2 | 12 |
| Answer-format content | FAQ blocks and concise, quotable sentences a model can lift verbatim | 12 | 10 | 11 |
| Server-rendered facts | Key facts in the initial HTML, readable without running JavaScript | 12 | 12 | 12 |
| llms.txt | An /llms.txt manifest declaring the site's purpose and key pages | 10 | 0 | 10 |
| Metadata | Unique title and description per page, canonical URL, hreflang, OpenGraph | 10 | 5 | 10 |
| Semantic HTML & headings | One h1, logical heading order, landmarks, descriptive links | 9 | 7 | 9 |
| Entity clarity | Consistent name, address and contact — a clear who, what and where | 8 | 4 | 7 |
| Accessibility | Alt text, ARIA, language and contrast — which also help machines read it | 7 | 5 | 7 |
| Freshness | Visible and structured last-updated dates; sitemap lastmod | 5 | 1 | 4 |
| Total | 100 | 46 | 96 |
We scored our sister company justidosolutions.com the same way and got about 83 — independently reconstructing its published 84/100. That tells us the rubric is honest, not flattering. Then we ran the same fixes on justidosolutions.com itself: it now scores 98.
How we got from 46 to 96
- Structured data (0 → 14): Organization and WebSite schema across the whole site, plus FAQ, Service and founder Person data — all cross-linked so a model reads one connected entity.
- AI-crawler access (2 → 12): a robots file that welcomes the AI crawlers by name, and a full sitemap with language alternates.
- llms.txt (0 → 10): a plain-language manifest at /llms.txt telling AI tools what we do and where the key pages are, in English and German.
- Metadata (5 → 10): a canonical URL and hreflang on every page, and proper page-specific social cards.
- Entity clarity (4 → 7): both our legal entities and our contact details now sit in the structured data.
- Freshness (1 → 4): real last-modified dates in the sitemap — and now inside the structured data too.
- Semantic headings (7 → 9): a demo mock-up on our homepage was adding a heading the page outline didn't need — cleaned up so machines read a tidy outline.
- Accessibility (5 → 7): an icon audit — every decorative icon is now invisible to screen readers and AI crawlers alike, so nothing muddies the text they read.
What we haven't finished
We'd rather show you this than hide it. A few signals are still short of full marks on our own site:
- Freshness (4/5): last-updated dates live in the sitemap and the structured data — visible on-page dates are the final step.
- Structured data (14/15) and answer-format content (11/12): the last point in each is polish — breadcrumbs everywhere and even more liftable phrasing.
- Entity clarity (7/8): one more pass so every mention of who we are and where we work lines up word for word.
How we measured
Each site is fetched as raw HTML — no JavaScript run — along with its robots.txt, sitemap and llms.txt, then scored signal by signal against the rubric above. The 'after' score is measured on the live, deployed site. Anyone can reproduce it by re-fetching the page and re-applying the same ten checks.