Run it on this page
Not a mockup. The engine is analyzing the page you are reading right now. Every finding below is real.
This marketing page scores poorly — and that is the point. If it scored 95, we would be lying.
Search changed.
Content teams didn't.
When someone asks ChatGPT, Perplexity, or Google's AI Overviews about your product, the answer is assembled from a handful of passages pulled from a handful of pages.
If your pages aren't structured for extraction, you don't appear.
SEO tools measure rankings. This measures something different: whether AI can quote the page.
Two surfaces.
One engine.
Pages Context Panel
Runs in the SitecoreAI Page Builder sidebar. Open a page — the panel returns a score, a prioritized list of fixes, evidence, and rewrite suggestions.
- Auto-analyzes on edit
- Priority fix queue
- Severity filter with counts
- Source tags on every finding
Fullscreen Extension
A site-wide view with every navigation page in a tree and scores populated as pages are analyzed. A priority list ranks the pages that need attention most.
- Page tree with score dots
- Site average and distribution
- Priority improvements queue
- Batch analysis with progress
Seven categories.
One hundred points.
Heading hierarchy, opening paragraphs, question-style headings, scannable blocks
Self-containment, pronoun density, context-dependent references
Numbers with units, dates, comparisons, concrete claims
Definitions, title alignment, author signals, metadata hygiene
FAQPage schema, visible Q&A content, schema-to-content consistency
dateModified, article:modified_time, ISO 8601 format, staleness threshold
External source links, named attributions, blockquote cite attributes
Standards-based detection. Editorial scoring, disclosed in-app.
Built for teams
who need to trust the number.
Deterministic
No model produces the score. Auditable. Reproducible. Same page, same number, every time.
Evidence-first
Every finding carries the exact sample that triggered it. Heading text. Paragraph excerpt. Missing property.
Disclosed
Findings backed by a published standard say so. Editorial rules say "heuristic." The user is never misled.
Page Builder native
Runs where the marketer works. Same engine as the site-wide view. No second system to trust.
Honest about limits
Not a ranking score. Not a citation guarantee. Published HTML only. No LLM in the pipeline.
Frequently
asked.
No. SEO tools measure ranking position. This measures whether AI answer engines can extract passages and cite your page. Both signals matter — they just answer different questions.
No. The engine is a deterministic rule pipeline. No LLM, classifier, or trained model is used. Re-running the same published HTML against the same analysis timestamp and policy produces the same score.
Standards-based findings parse published specs — Open Graph Protocol, schema.org, HTML Living Standard, Google crawler docs. Heuristic findings use editorial rules we chose. Both appear in the report, and every finding is tagged with its source.
No. The app reads published HTML and produces findings. It does not write to Sitecore content. Rewrites are shown as before/after suggestions in the panel — you decide what to apply.
Because the Sitecore starter kit does not publish a dateModified or article:modified_time on page templates. Adding one meta tag to your layout fixes it across every page.
The engine analyzes published HTML. Unsaved edits in Page Builder are not reflected until you publish. This is deliberate — it ensures what AI crawlers actually see is what gets scored.