Thin Content, and When the Word Count Is Wrong

Our thin content check told us youtube.com had 234 words. The other half of the same problem is a page with plenty of words and no links between them. The same record carried 44,704 characters of text, which works out at 191 characters per word. English averages about six.

What Google says it rewards, rather than what a word count says, is set out in its guidance on helpful content.

That is not a thin page. It is a page the counter could not read, and the finding fired anyway.

The number to check first

Every report shows two figures in the SEO text block, the word count and the text length. Divide one by the other. A real page lands between five and seven characters per word, because that is what written English does.

Where the result is thirty or more, the word count is not measuring your page. The two numbers disagree with each other, and the word count is the one driving the finding.

A page we can check by hand makes the case. Our own report carries www.iknowit.co.za at 558 words against 64,367 characters, a ratio of 115 to 1. Fetching that page directly returns 2,474 characters of visible text and 372 words, at 6.7 characters per word. The stored figure is roughly twenty-six times the real text length.

Why the count can be wrong

The crawl reads the page as a browser would and extracts the text, and pages that build themselves in JavaScript, or that render inside a canvas, or that place most of their copy in a component the extractor does not descend into, come back with less than a reader sees.

The failure runs the other way just as often. Where the extractor picks up markup it should have discarded, the character count inflates while the word count stays close to the truth, and the two figures separate.

Either way the finding is describing the extractor, not the page you published. Writing more copy will not change what the next crawl reads, and the report will keep raising it.

What thin content actually means

The check has a real job. A page carrying two sentences and a form gives a search engine almost nothing to work with, and it rarely answers the question somebody typed to arrive there. The finding is rated low and costs five points, because the exception is common enough to matter. A landing page whose intent is transactional can rank on very little, and the finding's own text says to judge it against what the pages competing for the term provide.

That comparison is the useful part and no word count substitutes for it. Open the three pages ranking above you for the term and read them. If they run to a thousand words and yours runs to two hundred, that gap is real. If they are product pages of eighty words and yours is a product page of eighty words, the count was never the problem.

What to do when the finding is right

Add the thing a reader came for, not the thing that fills space. The three sections a thin service page usually lacks are what the service includes, what it does not include, and what happens after somebody gets in touch. Those are specific to your business, and no template supplies them.

Where the page is a stub for something covered better elsewhere, the honest answer is to remove it or fold it into the page that covers it. A thin page that exists to hold a keyword is a page a reader bounces out of, and the crawl that flagged it is telling you something true.

Where the page is long enough already, fix the extractor instead. If your content arrives through client-side rendering, serving the same text in the HTML means the next crawl reads what a reader reads. That is a change to how the page is delivered, not to what it says, and it corrects the count on every future test.

The two figures are worth watching together

If you only ever look at the word count, a page like youtube.com reads as one of the thinnest on the web. The text length beside it says otherwise, and the pair is what makes the mistake visible.

We display both for that reason. Where they disagree by a wide margin, read the page yourself before you write anything. The guide on missing meta descriptions covers the closest relative of this finding, which is the other place a page's own text is measured and quoted back to you.

The fix, on each kind of thin page

Thin is a word count against a page's job. A contact page at 200 words is complete. A product page at 200 words is empty.

A real page: answer the next question Length is a symptom. The cause is that the page answered the question the visitor typed and stopped there.
// before: what the page already says
We offer roof repairs and insurance work.

// after: the questions a visitor arrives with
We offer roof repairs and insurance work across the Garden Route.
- What a standard assessment covers and what it costs
- How long a reinstatement takes, and who speaks to the insurer
- What happens if it rains while the roof is open
A genuinely thin page: merge or delete Google treats a set of thin pages as a signal about the site, not just the page. Removing them is a fix, not a loss.
# if the page has no more to say, it is not a page
# two 200 word pages on the same subject -> one 400 word page
# a page that exists for a menu item -> a section on the parent page
When the content is a widget A page that renders entirely from JavaScript is thin to anything that does not run it. The noscript text is what a crawler reads.
<noscript>
  <p>Enter your address to see the assessment price. The full price list is available as a <a href="/pricing/">downloadable PDF</a>.</p>
</noscript>