Thin content is not just short content. Here is how to find the pages Google crawls but refuses to index, and what to do with each one.
October 22, 2026 · Updated October 22, 2026 · 9 min read
By Daniel Petrica · I build the scanner and audit sites with it every week.
Educational content only. This guide explains what Google has publicly documented about how search works. Following these practices does not guarantee any specific ranking position. Google uses hundreds of ranking signals, and no single factor determines where your site appears in search results.
Thin content is one of the few SEO problems that gets worse the more you publish. If a site has a hundred pages that do not deserve to rank, adding another twenty does not dilute the problem — it confirms it.
This guide is the process I use to find thin pages, decide what to do with each one, and rebuild the pages worth saving. It is written for client sites, because that is where the damage is usually concentrated: a template gets published at scale, every URL looks slightly different, and nobody notices until traffic disappears.
Word count is the lazy definition. A 300-word page that answers a precise question can be useful. A 1,200-word page that rewrites the same five paragraphs as forty other pages can be thin.
The practical definition is simpler: a page is thin when a reader would get more value from a different page on the same site, or from the search result itself.
Google gives you a direct signal to work with. In Search Console, open Indexing → Pages and look at:
For a small site, "Crawled – currently not indexed" on a handful of pages is normal. When it is the majority of the site, stop publishing and start auditing.
You cannot fix what you cannot see. Pull three lists:
Merge them into one spreadsheet. For each URL, record:
| Column | Why it matters | |---|---| | URL and template | Group by template, not by URL. One bad template can create hundreds of thin pages. | | Index status | Indexed, crawled-not-indexed, discovered, redirect, 404, canonical conflict. | | Body word count | Count the main content only. Navigation and footer copy do not count. | | Primary query | The one search intent the page is supposed to satisfy. If there is no answer, that is a finding. | | Impressions and clicks (90 days) | A page with zero impressions after 90 days needs a reason to exist. | | Internal links in | Orphan pages are usually pages nobody needed. | | Last meaningful update | A date change is not an update. Did the content change? | | Unique asset | Original data, a test, a screenshot, a template, or first-hand experience. "None" is a finding. |
If the inventory is large, sample one URL per template first. Fixing a template is usually a content and engineering decision, not a hundred individual edits.
For each page, answer five questions:
A useful rule: one page, one intent. If a page tries to answer three different queries, it usually answers none of them well.
Do not start rewriting everything. Put each URL into one of five buckets:
The order matters: remove and merge before you create. Otherwise you are adding pages to a site that is already carrying too many.
For a "keep" or "improve" page, the fix is rarely "add 500 words". It is one or more of these:
For a "merge", the redirect is not enough on its own. The destination page must absorb the useful content from the pages you removed. A 301 to a page that does not mention the old topic is a soft 404 in practice.
A site published 90 comparison pages on a single day. Each page was 850–1,000 words and followed the same outline: where the competitor wins, where the site wins, a real example, a verdict table, and an FAQ. The pages were not copied, but they were templated. Many compared the product to tools that were not alternatives at all — infrastructure crawlers, free validators, local-listing platforms.
Search Console told the story:
The fix was not "write more pages". It was:
That is the shape of most thin-content recoveries: fewer URLs, more evidence, and a lot of patience.
Use this before publishing any new page, and when auditing an existing one:
There is no minimum that makes a page useful. As a working floor, a guide needs enough room to answer the query, show an example, and handle the obvious follow-up questions. In practice that is usually 1,000–1,500 words. A comparison page can be shorter if it has real testing behind it. Word count is a symptom, not the diagnosis.
Delete or redirect pages with no links and no demand. A 301 to the closest relevant page is better than a noindex because it consolidates signals. Use noindex only when the page must stay live for users but should not appear in search — for example, a filtered view or a member-only page.
Deindexation can happen in days. Reindexation after a site-wide quality problem usually takes weeks to months, and it depends on Google recrawling the cleaned-up pages. Watch Crawled – currently not indexed and the indexed-page count in Search Console. Those move before traffic does.
You can draft with whatever tool you like. The problem is not the tool; it is publishing pages that have no reason to exist. If a rewrite does not add first-hand evidence or a unique angle, it is still thin, and Google will treat it the same way.
Run the checklist against your own site. If the inventory is too large to do by hand, start with the pages Search Console already refuses to index.
Run a free scan to see how your pages score, or work through the 25-point SEO audit checklist if you want the full process.
See how your pages score against the checks in this guide. Drop in a domain, get a full scored report in minutes. No account needed to start.
Run a free scan →SEO audits
A practical, step-by-step SEO audit checklist for freelancers and agencies. Every on-page and technical factor worth checking, based on Google Search Central documentation.
SEO audits
An SEO audit report turns a crawl into a 0-100 score, an executive summary, and a prioritized list of issues. Here's exactly what's inside, with a live example.