Check format coverage before publishing
Audit whether a draft includes the useful content formats its topic and intent call for.
Content Completeness Audit
Free, no signup. See which useful content formats are already substantive, which are thin, which the page genuinely warrants—and which would only pad it.
Audit one page
Do not submit confidential drafts, signed/private URLs, passwords, API keys, access tokens, personal data, or other secrets. A public URL may include a query for fetching, but its credentials, query, and fragment are removed from displayed, shared, and exported source metadata.
Your input is sent to the AI model to generate a result — it is not stored afterward. Editorial AI is strictly opt-in. With it off, paste analysis stays in this browser; URL mode still sends the entered public URL to the bounded page fetcher. Results are not stored by the audit endpoint. 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
Deterministic example — editorial AI remains off
Present and substantive: step-by-step instructions, anti-patterns, validation tests, and measurement guidance.
Not evaluated: absent formats whose editorial value cannot be established by structure alone.
How to use it
- Enter one public URL or paste one page’s text or HTML.
- Leave editorial AI unchecked for a local deterministic presence audit, or explicitly opt in for missing-element gating.
- Review grounded hooks for every thin or suggested element and the anti-padding list for formats that should stay absent.
- Export the ledger, tickets, or brief stubs, make reviewed edits, then rerun the same page.
Site passport Local context for this saved site
Local data
Saved targets, named lists, and recent check summaries remain only in this browser.
Element verdicts
the anti-padding list
Don’t add these
These formats are correctly absent based on the supplied page. Adding them without new evidence would create filler.
Evidence excerpts
Rate this tool
How it works
The 14-element taxonomy and decision rules are derived from the editorial rubric used to review 473 of my own articles. Deterministic structural detectors always run first. When explicitly enabled, one bounded editorial pass decides whether absent formats are justified. Every add or thin hook must be an exact substring of extracted page text or code rejects it.
Limits
The tool evaluates one supplied page and makes no competitor, ranking, factual-accuracy, originality, traffic, or search-demand claim. URL fetches are bounded to 2 MiB and do not render JavaScript. Editorial AI receives at most the first 40,000 extracted characters and may return not-evaluated states. A suggested format still requires human research and editorial judgment.
Frequently asked questions
Is this a topic-gap or ranking score?
No. It checks whether 14 editorial formats are substantive, thin, absent but justified, or correctly absent. It does not compare competitors, predict rankings, or reward adding more sections.
Does the audit require AI?
No. The default audit runs deterministic presence checks in your browser. Editorial AI is off until you select the opt-in checkbox; missing-element decisions remain not evaluated without it.
What is sent when I opt into editorial AI?
For pasted or unpublished content, the supplied content is sent to the Product Worker and up to 40,000 extracted characters are sent to Cloudflare Workers AI. URL mode fetches the public URL server-side. Do not use confidential drafts, signed or private URLs, secrets, or credentials.
Does URL mode render JavaScript?
No. It fetches a bounded public HTML response. Paste rendered HTML when important content exists only after client-side JavaScript runsTurning HTML, CSS, and JavaScript into the final visual page and DOM..
Why are some verdicts not evaluated?
Truncated input, an unavailable or disabled AI service, exhausted budgets, invalid model output, or choosing the deterministic-only mode can all preserve honest not-evaluated states instead of inventing recommendations.
Feature requests for Content Completeness 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
Find structural omissions
Identify missing definitions, examples, comparisons, steps, evidence, or other deterministic format signals.
Compare two editorial drafts
Run the same fourteen-format checklist across versions to see which gaps a revision actually closed.
Add optional editorial review
Enable the clearly labelled AI gate only when qualitative review is useful, while keeping deterministic checks visible.
Watch the full workflow
Content Completeness Audit walkthrough
Read the transcript
Content Completeness Audit
Complete content is not content with every possible section. I’ll show you how to audit fourteen useful formats, run the browser-only example, interpret substantive, thin, and not-evaluated verdicts, avoid padding, use the optional editorial gate responsibly, export next steps, and understand the tool’s limits.
Step 1
This audit checks fourteen editorial formats such as steps, troubleshooting, validation, and measurement. It counts verdicts instead of assigning zero to one hundred because completeness is not a number to maximize. It does not compare competitors, predict rankings, or reward adding sections.
Step 2
Use it before publishing a guide, during a refresh, after merging overlapping pages, or when a draft feels long but still unhelpful. It helps distinguish a format that exists substantively, one that is thin, and one whose value cannot yet be established.
Step 3
Select Try an example. It switches to paste mode, loads a migration checklist, leaves editorial A-I off, and runs deterministic presence checks locally. This is the safest default for drafts: missing-element decisions remain honestly not evaluated rather than being invented from structure alone.
Step 4
The summary counts substantive, thin, worth adding, correctly absent, and not-evaluated formats. The attribution line states which layer ran and why any decisions remain partial. Read these categories independently; a larger present count does not automatically mean a better page.
Step 5
Present and substantive means the detector found enough structure and page language for that format. Each card shows a grounded excerpt and reason. Confirm the excerpt really demonstrates useful depth; pattern matching can recognize form without proving accuracy or reader value.
Step 6
A thin verdict means the format appears but may need more usable detail. Begin with the quoted page hook, then research what is missing. Add examples, decision criteria, failure handling, or validation only when evidence and audience needs support them.
Step 7
Not evaluated is an intentional result. With editorial A-I off, deterministic structure cannot decide whether an absent format belongs. The same state can appear for truncated input or an unavailable model. Do not silently convert unknowns into recommendations.
Step 8
When optional editorial gating validates correctly absent formats, the anti-padding list explains what should stay out. In deterministic-only mode, it may contain no decisions. Never add a section merely because the taxonomy names it; filler makes the page longer without making it more complete.
Step 9
Optional editorial A-I is off by default. If enabled, pasted content goes to the Product Worker and up to forty-thousand extracted characters reach Cloudflare Workers A-I. Never submit secrets, private or signed U-R-Ls, credentials, personal data, or confidential drafts.
Step 10
Export the C-S-V ledger for comparison, development tickets for substantive follow-up, or grounded brief-section stubs when a format is genuinely worth adding. In partial mode, exports preserve the available verdicts; they do not manufacture additions from not-evaluated rows.
Step 11
The audit covers one supplied page and makes no factual-accuracy, originality, traffic, ranking, or demand claim. Public fetches stop at two mebibytes and do not render JavaScript. Editorial A-I sees at most forty-thousand extracted characters and may return honest not-evaluated states.
Step 12
Choose a thin or justified format with a grounded hook. Confirm the audience question, gather supportable facts, write only the useful section, and have a human review accuracy and fit. Keep correctly absent formats out, rerun the identical page, and compare changes through the exported ledger.
Add value—not section count.
Start with one grounded thin element, confirm what readers genuinely need, and improve it with accurate evidence. Leave unjustified formats absent. After human editorial review, rerun the same page and compare the ledger, remembering that useful completeness is a judgment—not a score to maximize.
Common issues & how to fix them
- Information Missing content formats were not evaluated Fix: Keep the deterministic presence results as-is, or explicitly enable editorial gating when you need a bounded decision about whether an absent format is justified.