A plain-text version of this site for AI assistants is at https://contengi.com/llms.txt, with the full product documentation in one file at https://contengi.com/llms-full.txt. Every page is also available as markdown by adding .md to its URL.

Free tool / Helpful content checker

Is your content actually helpful? Check it.

Score any article or draft against the questions Google’s own guidance asks about helpful content and spam, plus how human it reads and how easily AI engines can quote it. Every problem comes with the line that caused it and the fix.

Plain code, no AI model Same page, same score, every time Nothing you check is stored

Eight dimensions, 30-plus checks

Six dimensions come straight from Google’s questions on helpful content and its spam policies. Two are ours: whether it reads like a person wrote it, and whether AI engines can read and quote it.

Experience

Signs the writer did the thing: “we tested”, “our data”, screenshots and photos nobody else has, original research. Counts double, because it’s the hardest thing to fake.

first-handoriginal dataown images

Effort

Depth, subheadings a reader can scan, named companies and real numbers instead of generic advice, and formats beyond paragraphs.

depthspecificstableslists

Sourcing

Links to outside sources, statistics with nothing behind them (quoted back to you), and whether the page says when it was last updated.

outbound linksunsourced statslast updated

Trust

A real person’s byline (not “Admin”), an author page, who runs the site and how to reach them. On money, health, and legal topics, expert credentials too.

bylineauthor pageabout + contactYMYL

People-first

Hype in the headline, a title that promises one thing and an H1 that says another, intros that clear their throat for 400 words, walls of text, and ads.

headlineintroreadabilityads

Spam signals

Keyword stuffing, hidden links, sneaky redirects, and thin affiliate pages. We also fetch the page twice, as a visitor and as Googlebot, to catch cloaking.

keyword stuffinghidden linkscloaking

Human voice

The engine behind our AI slop cleaner: stock AI phrases, filler, “it’s not X, it’s Y”, invisible characters, and sentences that all run the same length.

slop scorestock phrasesrhythm

AI readability

Whether the article is in the HTML crawlers get, one clear H1, a clean heading outline, sections that open with a direct answer, and Article schema.

server HTMLanswer-firstschema

Why there’s no AI model in it

Most content graders hand your article to a language model and ask it for a number. Run it twice and the number moves. Ask why and you get a confident paragraph that may or may not match what’s on the page.

This checker is plain code. When it says a statistic has no source, it quotes the sentence. When it says the intro runs long, it counts the words. When it says nobody wrote the article, it’s because there’s no byline in the page or its schema.

The trade-off is honest: code can’t tell whether a claim is true, so we score whether it’s sourced. What you get back is a list of things you can fix today and check again tomorrow.

An AI grader says

This checker says

“The content could benefit from more authoritative sources to enhance credibility.”

Score: 72, then 78, then 69

3 of 5 claims unsourced

“70% of buyers read three to five articles before talking to sales.”

Score: 64, every time

FAQ

Common questions

Is this Google’s helpful content score?

No. Google doesn’t publish a helpfulness score for anyone to check. This tool turns the questions in Google’s “Creating helpful, reliable, people-first content” guide and its spam policies into checks you can see on the page. A high score means the page shows the signals that guidance asks about. It can’t promise a ranking.

Why doesn’t it use AI to judge the content?

Ask an AI model to rate the same article twice and you’ll often get two different scores, with no way to see why. Every check here is plain code that points at something on the page: a sentence, a link, a missing byline. Run the same page tomorrow and you get the same score, so you can tell whether your changes worked.

How is the score worked out?

Each check passes (full points), warns (half), or fails (none), and the bigger checks carry more weight. Dimension scores average into the overall score, with Experience counting double, Sourcing one and a half times, and AI readability half. Spam signals don’t add points for being clean. They only take points away.

What is a YMYL page?

Your money or your life: pages about finance, health, legal matters, or safety, where bad advice can hurt someone. Google holds them to a higher standard. When the checker spots one, Trust counts double and it looks for a qualified author or an expert review on the page.

Can I check a draft before it’s published?

Yes. Switch to “Paste a draft”. Markdown works best: # for the headline, ## for sections, and links written as [text](https://...) so the checker can see your sources. Checks that need a live page, like the byline, dates, schema, and cloaking, are skipped rather than failed. Pasted text is scored in your browser and never sent to us.

Why can’t it read my page?

Some firewalls turn away every automated request, and a few sites only let real browsers in. Paste the article text instead. If the checker says it found too little text, the address may be a home or category page rather than an article.

Is the page I check stored?

No. We fetch it, pull out the article, and send the result to your browser, which does the scoring. The only thing we log is your IP address and the time, so one person can’t run hundreds of checks an hour.

Helpful by default, not by audit.

Contengi writes from your knowledge base: your data, your customers, your point of view. So the first draft already has the experience, specifics, and sources this checker looks for.

Also free: the AI slop cleaner, the AI bot view, and the LLM visibility check.