Will an AI answer actually cite you?
AI Mode and AI Overviews pull individual passages, not whole pages. Give me a URL and I'll grade your intro paragraph against your core term, plus your FAQ questions against their answers if you want. Or paste your own headers and passages and I'll grade each pair. Runs in your browser. Nothing is stored.
The main topic the page targets. On the URL tab I grade your intro paragraph against this. Not needed when you paste your own headers and passages. Write it the way a person would search, for example "project management software".
On a URL I grade only your intro paragraph against the core term. Tick the box to also pull your FAQ section and grade each question against its answer. Navigation, forms, on-page tool widgets and everything else are ignored.
Paste each header on its own line with its passage underneath, and a blank line between pairs. I grade each passage against its own header, one box per pair. The core term is not used here.
The grader scores relevance with a sentence-embedding model that runs in your browser. The first run downloads the model once (about 25MB), so give it a few seconds. Scores are directional and meant for comparing passages, not for matching Google's exact numbers. See what it does and doesn't do below.
The grader finds the weak passages. We rewrite them to win.
Free shows you which passages are too weak to be cited. Our service does the rest. We pull the real sub-queries AI systems generate, rewrite each weak passage as a direct, source-backed answer, fact-check every stat against its original source, and hand back copy built to get retrieved, with placement notes for each change.
- Real fan-out sub-queries, not your best guess
- Weak passages rewritten as direct, citable answers
- Every stat verified against its original source
- Content gaps filled with new passages
- Ship-ready copy with placement notes per change
Send us a URL or doc
Prefer to talk first? Book a 30-min call →
A relevance meter, not a rewrite engine.
The free grader runs entirely in your browser and never touches Google. Here is exactly where the line sits.
The free tool does
- On a URL, grades your intro paragraph against the core term
- Optionally grades each FAQ question against its answer, ignoring the rest of the page
- On the paste tab, grades each header against its passage
- Scores relevance with in-browser sentence embeddings
- Flags weak passages, tells you why, and notes near-duplicates
- Hands you a prioritized, copyable list of what to fix
It deliberately does not
- Pull the real fan-out sub-queries Google generates
- Compare you against the passage currently winning the AI Overview
- Rewrite weak passages into citable copy
- Use Google's production embedding model (this one is directional)
- Iterate a rewrite until it beats the benchmark
- Guarantee a citation
About the grader.
What exactly does it grade?
On the URL tab it grades your intro paragraph against the core term you enter, and nothing else unless you tick the FAQ box. With that box ticked it also finds your FAQ section and grades each question against its answer. On the paste tab it grades each header you paste against the passage beneath it.
How does it find my intro and headings?
The intro is the first real body paragraph, not the headline and not navigation or tool copy. For FAQs it looks for a heading like "FAQ" or "Frequently asked questions", then pairs each question under it with the paragraph that answers it. Anything outside that FAQ block is left out.
What does the score actually measure?
The score is the cosine similarity between a target and a passage, computed from sentence embeddings in your browser. It estimates how semantically close a passage is to the topic it should answer. As a rough guide: 0.50 and up is strong, 0.40 to 0.50 is borderline, and below 0.40 is weak.
Is my content stored anywhere?
No. The scoring runs in your browser. When you grade a URL, our server fetches that page's body once and hands it back to your browser to score, with navigation, forms and on-page tool widgets stripped out. Nothing is logged or saved on our side.
How do I fix a weak passage?
Lead the passage with a direct answer that uses the target's exact words, state it as a clear subject-verb-object fact, add a specific number, name, or date, and keep it tight at two to five sentences. The grader flags which of these each weak passage is missing so you know where to start.
What's the paid version?
A done-for-you rewrite. We pull the real fan-out sub-queries, rewrite each weak passage as a direct citable answer, fact-check every stat against its original source, fill content gaps with new passages, and hand back ship-ready copy with placement notes. Send a URL or doc through the form above to start.