Preflight answer-focused content
Run a deterministic check on extractability, answer-ready structure, and citation trust before publishing.
AI Content Audit
Free, no signup. One pass over your content gives you a single GEO/AEO score built from three lenses — is it written to be quoted, shaped to be answered from, and sourced well enough to be trusted — plus a merged, prioritized fix list.
Before
A long article with no direct answer, unsupported statistics, and dense academic prose.
After
A page that opens with a direct answer, quotable sentences, and sourced claims — ready for AI retrieval and citation.
| Lens | Measures | Weight |
|---|---|---|
| Extractability | Readability grade and mean sentence quotability — can a passage stand alone when lifted? | 35% |
| Structure | Answer-ready citability — direct answers, chunk relevance, freshness, author signal | 40% |
| Trust | Citation coverage — unsupported claims, aging evidence, broken citations | 25% |
Pasted content is analyzed in your browser. A URL is sent only to the protected fetch endpoint; paste content takes priority. After a successful URL check, only its domain is remembered locally.
Checks run from our server; we fetch the URL you enter and don't keep the results. The three-lens audit runs in your browser. Up to 25 cited URLs may be sent to the bounded status checker; pasted prose is not sent. Anonymous run-level outcome counters may be used for aggregate research; URLs, domains, IPs, and identifiers are never included, and no statistic is released below 100 runs.
Sample report Built-in deterministic example
The Try an example button loads a short, attributed paragraph with one aging statistic. The current engine returns:
GEO/AEO score: a numeric 0–100 result
Extractability · Structure · Trust — each shown as a separate sub-score
✓ Direct answer near the start
✓ Visible update date
✓ Author or reviewer signal
One prioritized fix: review the aging 2021 statistic
How to use it
- Paste one page's visible text or HTML, or fetch a public URL.
- Select Run content audit, or use the built-in example to see the output shape.
- Start with the overall score and the three sub-score bars. Click a bar to expand its lens detail.
- Work through the prioritized fix list, highest-impact first. Re-run after changes to confirm the score improves.
| Lens | Weight | Signal |
|---|---|---|
| Extractability | 35% | Readability grade, mean sentence quotability |
| Structure | 40% | Chunk relevance, answer coverage, quotability density, direct answer, freshness, author |
| Trust | 25% | Citation gaps, aging evidence, broken citations |
Site passport Local context for this saved site
Local data
Saved targets, named lists, and recent check summaries remain only in this browser.
Rate this tool
What the results mean
- Overall GEO/AEO score — a weighted blend of the three sub-scores using the weights shown above. Higher is better, but the score is a heuristic, not a retrieval or citation prediction.
- Extractability — how readable and quotable the content is. Penalized for high grade level and low mean sentence quotability.
- Structure — how well the page gives an answer engine something clear to retrieve. Driven by chunk relevance, answer coverage, quotability density, and three content checks (direct answer, freshness, author).
- Trust — citation and evidence quality. Penalized for unsupported claims, aging statistics, and broken or questionable citations.
- Prioritized fix list — signals from all three lenses merged into one list, ranked by approximate score impact. Fixes that span two lenses (e.g. a buried answer that affects both extractability and structure) appear once with both tags.
How it works
The content extractor normalizes the input (HTML or plain text) into readable blocks. All three engines then run over the same extracted text: the readability scorer computes grade levels and sentence quotability; the answer-ready inspector checks chunk relevance, answer coverage, and three content patterns; the citation-gap analyzer finds unsupported claims and sends up to 25 cited URLs to the bounded status checker. Each lens normalizes its raw output into a 0–100 sub-score using the documented rules above, and the overall score is the weighted sum. The fix-list builder maps each lens signal to a fix entry, deduplicates overlapping signals, and ranks by impact.
Features
- Single 0–100 GEO/AEO score with all three sub-scores visible.
- Three expandable lens detail panels with the same depth as the standalone tools.
- Merged, impact-ranked fix list deduplicated across lenses.
- Text or HTML input; optional URL fetch via the shared
/api/text-fetchendpoint. - Citation link-status pass reuses the bounded
/api/check-statuschecker (up to 25 URLs). - CSV ticket export of all findings.
- Everything runs in the browser; no model is called.
Limitations
The readability and quotability engines are English-oriented and can misread names, abbreviations, and specialized vocabulary. Answer-ready checks detect simple wording patterns, not meaning. Citation adjacency is a crude proxy: a URL in the sentence may be irrelevant, while a footnote may be valid but missed. The three-year aging threshold is intentionally generic. The bounded status checker does not read papers, assess source quality, or determine whether a citation entails the exact claim. The overall score is a transparent heuristic blend — it cannot predict whether any specific AI model will retrieve or cite the page.
Deep-dive tools
Each lens is also available as a focused standalone tool:
- Quotable Sentence Finder — full readability and sentence-level quotability detail.
- AI Citation Readiness Checker — citability with optional retrieval evidence and per-model freshness view.
- Citation Gap Checker — full claim-level gap and aging-evidence report.
Related: Chunk Tester, Commodity Content Checker, On-Page SEO Checker.
Frequently asked questions
Is this an AI visibility score?
No. The score is a transparent weighted blend of three deterministic writing heuristics: readability/quotability, answer-ready structure, and citation coverageAn AI citation is the visible source link an AI answer engine shows next to its generated text — the clickable reference that credits the web page it used. A citation's presence is a separate thing from whether the cited page actually supports the statement, and from being retrieved (read behind the scenes) or merely mentioned (named without a link); citation is driven more by brand mentions and being retrievable than by traditional ranking.. It does not observe or predict any specific AI model, answer engine, or ranking system.
Why are the weights 35/40/25?
Structure carries the largest share because a well-shaped answer passage is the single strongest citability signal in a content-only audit. Extractability (readability and quotability) is second. Trust (evidence and citation quality) is third because a well-structured, readable page can still be cited without perfect sourcing, but not without structure.
Does the tool check my links?
In URL mode the page sends up to 25 cited URLs to the bounded status checker. A failed status can flag a broken citation; the tool does not read page content behind the link or verify that a reachable source supports the claim.
Can a high score guarantee an AI citation?
No. A high score means the content satisfies the three local heuristics. Whether an answer engine retrieves and citesAnswer Engine Optimization (AEO) is the practice of structuring content so engines deliver it as a direct answer — featured snippets, voice assistants, and AI search — rather than just a ranked link. Coined for voice search in 2018 and revived for the LLM era. Google's position is that it's still SEO. a page depends on external factors (rank, authority, indexation, model behavior) that this tool deliberately does not measure.
Why is a claim flagged when the sentence links to a source?
The tool checks whether a URL appears in the same sentence as a claim. A footnote, endnote, or nearby link may be valid but still missed by the adjacency heuristic. Always verify the cited source supports the specific claim.
Is pasted content uploaded?
No. The three-lens audit runs entirely in your browser. Only the optional URL fetch and the bounded status check (up to 25 URLs) involve a server call.
Feature requests for Ai Content Audit
Upvote what you want most. New ideas can be submitted from the floating Feedback menu; requests appear here once approved, and the most-wanted rise to the top.
You won't be emailed about that request anymore.
Loading…
➕ Request a feature
New requests are reviewed before they appear here.
Where this tool helps
Common use cases
Prioritize GEO and AEO edits
Use the transparent score breakdown to find the highest-impact structural and evidence gaps first.
Compare competing drafts
Audit two versions with the same rules to see which one presents clearer, better-supported answers.
Verify a revision
Rerun the updated copy and confirm which specific checks improved instead of relying on the total score alone.
Watch the full workflow
AI Content Audit walkthrough
Read the transcript
AI Content Audit
Content can be accurate yet difficult for readers and answer systems to extract, quote, or trust. I’ll show you when this audit helps, how to run an example, read every score and recommendation, understand its limits, and turn the results into a practical revision plan.
Step 1
This deterministic audit blends three visible lenses: extractability at thirty-five percent, structure at forty percent, and trust at twenty-five percent. It does not call an A-I model, measure search visibility, or predict whether an answer engine will retrieve or cite the page.
Step 2
Use it before publishing an important guide, while refreshing an aging page, after combining several writers’ work, or when a technically correct answer feels hard to scan. It is especially useful for finding missing direct answers, weak structure, stale statistics, and unsupported claims.
Step 3
Paste text, fetch a public U-R-L, or select Try an example. The example runs immediately and keeps the prose analysis in your browser. Public page fetching and bounded citation status checks are separate network actions, so review the privacy note before using sensitive material.
Step 4
The overall score is a weighted heuristic from zero to one hundred. Read it with the three subscores rather than as a grade. A strong total can still hide one weak lens, and a low score may reflect deliberate writing choices that are appropriate for the audience.
Step 5
Extractability checks whether sentences are readable and useful passages are easy to quote. Review sentence length, dense wording, and whether key ideas stand on their own. Simplify only where clarity improves; do not remove necessary nuance just to raise a number.
Step 6
Structure looks for answer-ready signals such as a direct definition, clear headings, an author, and an update date. Add these when they help a reader orient quickly. The tool recognizes simple patterns, so confirm that any detected signal is meaningful and not merely present.
Step 7
Trust identifies claim-like language, adjacent links, years, and evidence signals. An older statistic becomes a review prompt, not an automatic failure. Verify the original source, method, population, date, and whether it actually supports the exact wording before citing or revising the claim.
Step 8
The sentence map shows where signals appear in the analyzed passage. Use it to locate dense stretches, answer-like sentences, and claims needing evidence without losing their surrounding context. A highlighted sentence is an observed pattern, not proof that the sentence is correct or incorrect.
Step 9
Prioritized fixes translate the lens findings into an editorial queue. Start with changes that improve both reader comprehension and evidentiary quality. Export the report if you need to assign work, then document what changed and why rather than chasing a perfect score.
Step 10
Features include pasted text and public U-R-L input, a weighted score with separate subscores, expandable lens details, a sentence map, prioritized fixes, and export. When citations are present, up to twenty-five U-R-Ls can receive bounded status checks without sending the whole prose for model analysis.
Step 11
The patterns are English-oriented and inspect wording more than meaning. Citation proximity is crude, the three-year aging rule is generic, and a reachable link says nothing about source quality or entailment. The audit cannot predict retrieval, ranking, citation, factual accuracy, or audience satisfaction.
Step 12
Choose one weak lens, revise the relevant passages, verify facts and sources, then rerun to compare. Read the result aloud, test the main question with real users or search data, and open the deep-dive tools for quotability, citations, or other specific issues that need a narrower workflow.
Use the score to prioritize—not to predict.
Start with the highest-impact fix, preserve factual accuracy, add direct evidence where it genuinely supports the wording, and rerun the audit. Then read the page yourself, test important queries, and use the deeper tools when a specific extractability, structure, or citation issue needs closer investigation.
Common issues & how to fix them
- Warning A dated claim needs fresher supporting evidence Fix: Replace or contextualize the aging statistic with a current primary source, preserving the source date and scope beside the claim.