Summarize this article with AI
An empty category page here, a stubby article there, and a list of excluded URLs that just keeps going. Which of them really needs rewriting? Before you add a word or delete a page, a thin content SEO audit needs a decision you'd be comfortable defending.
Some pages need a better answer; other URLs need a different indexing decision. This workflow sorts one from the other. Along the way there's a quality checklist, two Search Console review queues, and a worksheet for picking the right change and confirming it worked.
Thin content SEO: what counts as thin content
Put simply, a page is thin when the person using it gets little or nothing of value out of it. Google's manual action for thin content covers low-quality or shallow pages, including thin affiliate pages, scraped content and doorways.1 Word counts can match and the right call can still differ: a short answer isn't an empty template, even at the same length.
Triage starts in Google Search Console's Manual actions report. If there's an action listed, the affected content and URL patterns jump ahead of any routine performance audit. No action? Then you work two queues, one for submitted pages that aren't indexed and one for indexed content that's losing visibility.
Key takeaways:
- Original value and depth matter; a word-count target doesn't.
- An indexing reason tells you where to look. It doesn't prove the content is thin.
- Indexed content gets reviewed in the Performance report, apart from the excluded URLs.
- Usefulness and relevance decide between improving, consolidating, redirecting, noindexing, or deleting.
- Measure the deployed change and its search results separately.
Why word count is not the test
Google Search Central says it has no preferred word count for pages.2 What its content self-assessment asks instead is if the topic gets substantial, complete coverage. A short page might point to missing depth. Padding it with filler won't add any.
Yoast SEO's word-count check recommends more than 300 words for a post or page.3 That is a plugin recommendation, not Google's thin-content threshold. Yoast itself says more words alone won't make content rank.
So word count is a clue at best. It's never the actual problem. If a short FAQ answer settles the question completely, it isn't thin. Flip it around, and a long product page made of nothing but the manufacturer's spec sheet is thin however many words it runs to.
Is the page thin? Ask these five questions
Go through the prompts below for each page, noting whatever's missing as you go. Google's helpful-content questions are the source for the first four. The fifth helps you decide what to edit. It's for planning edits, not for scoring rankings.
| Review question | Evidence to record | Next action if the answer is weak |
|---|---|---|
| Does it add original information, research, reporting or analysis? | The specific contribution that belongs to this content | Add an example, comparison, test or analysis you can support |
| Does it cover the topic at the depth the searcher needs? | The unanswered question or missing step | Complete the answer without expanding into unrelated topics |
| Does it add value beyond the sources it draws on? | What readers gain beyond a summary | Replace paraphrases with explanation or original work |
| Does it offer substantial value compared with search results? | A useful difference from other results | Improve the contribution or reconsider a standalone article |
| Can a visitor finish the task this content promises? | The decision or action the visitor can take | Rewrite around search intent and a clear next step |
Here's the subtraction test. Take out everything competing results already explain better and see what's left. Nothing useful left? Then you need a stronger contribution, and more copy won't be it. Ask someone outside your website team to critique whether the content helps them, as Google recommends for thin-content reviews.1
Say it's an affiliate headphone review. Original testing, ratings, or product comparisons would count; one more rewording of the merchant's description wouldn't. And a location template? Figure out what a visitor in that city needs to know before you ship yet another variant.
Copy this brief into the content audit:
URL or template:
Search intent and visitor task:
Original contribution:
Missing information or depth:
Overlap with existing content:
Proposed outcome and reason:
Evidence needed before editing:
Scaled content abuse, commodity content, and thin content in 2026
AI use alone does not make content thin. Google's scaled content abuse policy addresses many pages created primarily to manipulate search rankings without helping users. It applies to automated, human-written and mixed production methods.4
Research and structuring original content are things Google says generative AI can help with. Its condition is that a human fact-checks and reviews anything AI-generated before it goes out. Accuracy first, then contribution: those are the two tests I'd put an AI-assisted draft through.
What changed since Panda
Panda goes back to February 2011, when Google announced an algorithm update that went after low-quality sites: little added value, copied content, limited usefulness. In 2015 Panda became part of the core ranking systems. The helpful content system joined those systems in March 2024. Google reported 45% less low-quality, unoriginal content in search results after the March 2024 changes.4
Google's May 2026 guide on optimizing for generative AI search uses a term worth borrowing: commodity content, meaning material often built on common knowledge that usually adds little unique insight. It contrasts that with perspectives grounded in expertise or experience, such as a firsthand review instead of a summary of existing content.5 Raters, per Google's quality rater guidelines, give the Lowest rating to main content with little to no effort, originality, or added value behind it. Filler that makes a page look rich gets a warning there too. Those ratings don't change rankings directly, though. They're how Google checks its ranking systems.
This matters when you're commissioning content. What can this author bring that isn't just familiar tips in a new order? Google's guidance warns that creating separate content for search variations or fan-out queries primarily to manipulate rankings or AI responses violates its scaled content abuse policy.5
Types of thin content: spam patterns versus ordinary problems
Separate spam patterns from ordinary content issues such as internal duplication and underdeveloped templates. Google is fairly relaxed about duplication within a site: some is normal, and it isn't a spam-policy violation. That doesn't make it harmless. Users can find it frustrating, and it muddies your view of how content performs in search.
| Pattern | What to review | Practical response |
|---|---|---|
| Thin affiliate content | Reviews or descriptions copied from a merchant without original value | Add supported testing, comparisons or useful features |
| Scraped content | Republishing or lightly changing others' content without added value | Replace copying with an original contribution |
| Doorway pages | Similar query or city pages funneling visitors to a less useful intermediate destination | Check whether each destination serves a distinct visitor need |
| Scaled content abuse | Large amounts of unoriginal content produced primarily to manipulate rankings, including pages that make little sense to a reader but contain search keywords | Reassess the production approach and exclude abusive content from Search |
| Internal duplicates | Equivalent content at different URLs | Fix canonicalization or consolidate |
| Empty or underdeveloped templates | Archives, listings or variants that do not deliver the promised content | Improve the template or choose another outcome |
Now picture city pages where only the city name changes and every visitor gets funneled to one destination. That's Google's doorway pattern. For scraped content, attribution alone does not supply added value.2 In both cases, focus on the benefit to visitors rather than a surface difference in keywords.
Does thin content hurt your whole site?
Google's ranking systems work page by page, Google says, though site-wide signals and classifiers feed in too. Weak site-wide signals don't drag every page on the site down with them. So look at the affected section, and don't assume one weak article explains a drop across the whole site.
Widespread, clearly manipulative thin content (doorway pages, scraped content at scale) can trigger a manual action, not just a ranking drop.1 With a manual action, a human reviewer at Google has found the site breaking its spam policies. The report can identify affected URL patterns or all pages.
Crawl efficiency is a separate concern, and Google scopes it narrowly. Google's crawl budget guide is written for sites with 1 million or more unique pages whose content changes about weekly. It also covers sites with 10,000 or more unique pages changing daily, and sites with a large share of URLs in “Discovered - currently not indexed.” Google's crawl budget guidance says excessive crawling of unwanted URLs can keep its crawlers from exploring the rest of a site.
Google Analytics 4 can add user experience context, as long as you read its bounce rate correctly. There it's the share of sessions that weren't engaged, a number Search Console doesn't report at all. Compare similar pages, and let engagement raise questions about the content rather than prove it's thin.
How to find candidates using Google Search Console
Keep indexing and search performance in separate queues. An excluded URL's lack of impressions is not evidence that its content is thin. An indexed article with declining clicks needs a closer look at its content, not an automatic deletion.
I have diagnosed indexing, crawling and canonicalization issues while scaling programmatic SEO pages and large content sites. I group candidates by template, then flag pages that need a different outcome. That connects the audit to how the website produces pages.
Review submitted pages that are not indexed
Pull the Pages report.6 Pick “All submitted pages” or a single sitemap in the Page indexing report, and open “Why pages aren't indexed.” Don't lean on the examples alone, since that table cuts off at 1,000 rows; the counts and template patterns fill in the rest.
Should this group even be in Search? Answer that before you look at its content. Pages that aren't indexed are not necessarily errors, and duplicate or alternate pages can be correctly excluded.6
| Search Console status | What it tells you | Next investigation |
|---|---|---|
| “Crawled - currently not indexed” | Google crawled it but has not indexed it; it may or may not be indexed later | Check usefulness and duplication with the “Crawled - currently not indexed” workflow |
| “Discovered - currently not indexed” | Google knows the URL but has not crawled it; Google says crawling was typically rescheduled to avoid overloading the site | Investigate crawling before a content verdict with the “Discovered - currently not indexed” guide |
| “Duplicate without user-selected canonical” | Google treats it as a duplicate; this can be working as intended | Compare the content and preferred URL using the duplicate canonical diagnosis |
| “Soft 404” | Content suggests an empty page or an error without the appropriate response | Decide whether to restore useful content or remove it with the soft 404 decision workflow |
| “URL marked 'noindex'” | The URL is marked to stay out of Search | Confirm that exclusion is deliberate |
| “Not found (404)” | The URL is missing | Check whether it was removed without a replacement; that can be correct |
Review indexed content with weak or falling performance
Next comes the Pages view of the Performance report. Set comparable date ranges. What you're shortlisting is indexed content with impressions and few clicks, or a line that's heading down. Three months is the default view; you can go back 16.
Use weekly or monthly views to smooth daily changes.7 Suspect two pages overlap? A query filter will show you which pages turn up for it. If they look alike, compare what each page is meant to do. They don't prove a ranking problem.
Every candidate gets two notes: its search intent, and the information it lacks. The ones worth keeping on the list are posts or landing pages where you can name a concrete improvement or merge. Low traffic, by itself, says nothing about quality.
Group candidates by template before editing
Group the audit by archive, location, product variant or blog section. When every example has the same weakness, you're looking at a template change, with the exceptions listed separately.
Both queues fit in this one worksheet; just write the performance date range down. If traffic won't change the decision, skip that column. Record deployment dates rather than copying example dates into the audit.
| URL or pattern | Queue and status | Search date range | Clicks or trend | Content gap | Backlinks to review | Chosen outcome | Change date | Verification |
|---|---|---|---|---|---|---|---|---|
| [URL or template] | [Indexing reason or Performance] | [Start to end] | [Observation] | [Missing value] | [Relevant links] | [Decision] | [Deployment date] | [Evidence] |
How to fix thin content: choose among five outcomes
What users need, what's relevant, and whether the content has to stay available decide the outcome. Retiring a URL? Check three things first: traffic, backlinks, and the page you'd send it to. Deciding the content's fate is one job. Crawl control is another.
| Situation | Content outcome | What to verify |
|---|---|---|
| A distinct visitor need exists, but the answer lacks depth or original value | Expand | The revised content supplies the missing contribution |
| Several articles cover the same need and belong in one resource | Consolidate | The survivor includes useful material from the retired versions |
| The old content is retiring and a relevant replacement exists | Redirect | The destination answers the old visitor's need |
| Content must stay available to users but should be absent from Search | Noindex | Google can crawl the page and read the indexing rule |
| Content has no useful role or relevant replacement | Delete | The response matches permanent removal |
Expand it with an original contribution
Supply the information missing from your audit brief. Original analysis could fill that gap, or firsthand examples, product testing, or a comparison that helps visitors choose. While you're in there, cut the filler and anything inaccurate.
Does the topic even deserve its own resource? Check before you expand. A complete short answer shouldn't get stretched just to satisfy a plugin's word count.
Merge pages that overlap
Got several pages chasing the same search intent and the same keywords? Pick one survivor that actually satisfies that intent. Merge them into one page that actually covers the topic, then 301 the merged URLs to the survivor.8 Bring over the details that help readers; stacking several introductions together doesn't count.
Canonicalization, per Google, can pull signals like links together onto one preferred URL. When equivalent versions have to stay reachable, pick the canonical with canonicalization, not noindex. The alternate page with proper canonical tag guide covers that separate indexing outcome.
Redirect to a relevant replacement
Redirects fit when the retired content has a relevant replacement. Make sure the destination does the old page's job before you map anything. Google warns that sending many old URLs to one irrelevant destination can confuse users and may be treated as a soft 404.8
Write the mapping down and fix internal links in the same change. Our 301 versus 302 redirect workflow covers setting the redirect up and checking it. Backlinks are a reason to look harder at a URL. They're no excuse for forcing an unrelated destination.
Noindex content that must remain available
Noindex is for content users still need but Search doesn't, like thank-you or checkout pages. Keep the page live, keep it out of the index.9 Once Googlebot crawls the page and reads the tag or header, Google drops it from Search results.
Blocking it in robots.txt at the same time would defeat the noindex. Google must be able to crawl it to see the rule.9 Crawl control for unwanted faceted URLs is its own implementation choice.
| Control | Purpose and behavior | Implementation check |
|---|---|---|
| Noindex | Removes content from Search once Google crawls and reads the rule; crawling still occurs | Allow crawling so Google can see it |
| robots.txt | Blocks crawling; a blocked URL can still be indexed | For unwanted faceted navigation, follow Google's crawl-blocking guidance |
| 404 or 410 | Reports permanently removed content | Use when removal has no replacement |
For unwanted faceted navigation URLs, Google recommends disallowing crawling in robots.txt. Noindex won't save crawl budget, its guide says, because Google keeps requesting those URLs anyway. If an already indexed URL needs removal, Google advises removing the robots.txt block and using noindex; see “Indexed, though blocked by robots.txt”.6
Delete content without a useful role or replacement
Check search traffic, backlinks, and internal links before deleting anything. Be sure that improving or consolidating it wouldn't do users any good. Let it return a real 404 or 410 rather than redirecting it somewhere unrelated just to avoid an error.8
A “Not found (404)” entry can be correct when content was removed without a replacement.6 Record the removal as the chosen outcome in the worksheet. Do not reopen the content project simply to make that report entry disappear.
Copy this implementation ticket after choosing an outcome:
Affected URLs or template:
Visitor need being preserved:
Chosen outcome and reason:
Content additions, survivor or redirect destination:
Expected response or indexing rule:
Internal links to update:
Crawl access needed:
Deployment date and annotation:
Verification evidence:
How to measure whether the cleanup worked
Verify the implementation first, then compare its indexing and performance results. A working redirect, a noindex rule Google can read and a more useful article each need different evidence. Use the chosen outcome to define success before reviewing traffic.
Copy these checks into the verification field:
- Expanded content: confirm the missing answer or evidence is present.
- Consolidation or redirect: test the mapping and check that the destination lets visitors complete the same task.
- Noindex: the rule should be in place, and crawling should still be open.
- Deletion: expect a 404 or a 410 from the URL.
- Search results: record the indexing outcome separately from the click trend.
Check the change and indexing outcome
Test affected examples and exceptions after deployment. The Page indexing report's Validate fix feature checks all instances of the issue unless you filter to a specific sitemap. Google says validation stops when it finds a remaining instance, and typically takes up to about two weeks, though it can take much longer.6
Issue counts update when Google crawls pages with known issues, even without a validation request.6 Keep the content audit notes alongside those indexing counts so you can assess both outcomes.
Dealing with a thin-content manual action? Every issue on the affected pages has to be fixed before you click Request Review. Google says fixing only some pages does not earn a partial return to search results.1 In the request, include examples of what bad content went and what good content came in. Don't hide the affected pages behind a login, a robots.txt block, or noindex while you wait.
Google says reconsideration reviews can take several days or weeks.1 That timeline has nothing to do with your search-performance comparison, so keep them apart. Reconsideration is a recovery step after a manual action. You don't send one because you edited an article.
Compare search performance after the change
A right-click on the Performance report chart gives you “Add annotation,” which is how you pin the deployment date. Custom annotations are visible to people with access to the property.10 Do the same for content changes, template updates, and migrations. You'll want that context in later comparisons.
Match your date ranges, and look at the trend weekly or monthly rather than day by day. With consolidated content, look at the survivor and the retired pages together. Judging only the retired pages' traffic misses the point. The question to answer is narrow: after the change, did the resource gain useful search visibility?
How to prevent thin content from returning
What does this content add for users? Ask before you publish it or scale the template. I look for two things: a distinct visitor need, and enough substance for someone to finish the task. If an existing article could hold the idea, improve that article instead of starting a new one.
Google warns that adding or removing lots of content primarily to make a website appear fresh does not improve search rankings.2 It also questions changing article dates without substantial changes. A refresh should fix missing value or bad information. A date on the calendar isn't a reason.
Build these checks into publishing and template changes:
- State the search intent and contribution before drafting.
- AI-assisted content gets checked for accuracy and real added value.
- Examine populated and empty template examples before scaling.
- Record content and infrastructure changes with dated annotations.
- Come back to the affected indexing groups and performance trends once changes ship.
For the next audit, I'd open SEO Heatmap and compare every page's Search Console clicks, by month across 16 months or by week across 16 weeks. The AI insights pick out top gainers and losers, and those make a good shortlist to hold up against your worksheet's deployment notes. Then follow up on the groups still open and the edits you shipped. No need to rebuild the queue from zero.
Frequently asked questions
What does Google mean by thin content?
When Google issues a thin-content manual action, it's talking about low-quality or shallow pages with little or no added value. Its examples: thin affiliate pages, scraped content, doorways. Length alone won't tell you. Originality and how fully the answer covers the question will.
Is there a minimum word count for avoiding thin content?
Google has no preferred word count for pages. Yoast SEO wants more than 300 words on a post or page, but that's the plugin talking, not Google. Short and complete can be exactly what a reader needs. A long description lifted from a merchant can still add nothing.
Is AI-written content automatically thin?
No. Using AI doesn't, by itself, make content thin. Google's scaled content abuse policy targets lots of pages made mainly to game rankings without helping anyone, and it doesn't care if a human or a machine wrote them. Fact-check anything AI helped write, then ask what it offers that existing pages don't.
Why doesn't a list of excluded URLs prove I have thin content?
A page that isn't indexed isn't automatically an error. Duplicates and alternate versions can be excluded for good reason. Decide which submitted URLs should be in Search first. The indexing reasons come second. Audit indexed content with weak or falling visibility separately in the Performance report.
How is thin content different from duplicate content?
Thin content is about missing value, while duplicate content is about the same material living at another URL. Some internal duplication is just normal life on a website, and it's not a spam-policy violation. Before rewriting or merging anything, ask two separate questions: is it useful, and is the canonical right?
Should I delete thin content or use noindex?
Use noindex if users still need the page but Search doesn't. Delete when the content has no useful role and nothing relevant to replace it, and only after checking traffic, backlinks, and internal links. If it's gone for good, return a 404 or 410. And keep noindexed URLs crawlable. Otherwise the rule goes unread.
Does Google penalize thin content?
There are two separate mechanisms. One is the ranking systems, which use page-level and site-wide signals to favor original, helpful content. Second, there's the manual action, where a human reviewer finds spam-policy violations and the Manual actions report names the affected URL patterns, or all pages. Fix every issue on the affected pages before you hit Request Review. Fixing some of them doesn't get you a partial return.
How long should I wait to judge a thin-content cleanup?
Use deployment evidence and comparable search date ranges rather than a fixed ranking-recovery deadline. Validation in the Page indexing report usually takes up to about two weeks, and sometimes much longer. Reconsideration after a manual action runs several days to several weeks. Put an annotation on the change and track indexing and search trends apart from each other.
Would you rather catch slipping pages before they end up in next quarter's audit queue? SEO Heatmap turns your Search Console data into a heatmap of every page's clicks over time. It's free for one Search Console property.
Sources (10)
SEO Heatmap Knowledge relies on primary sources: Google's documentation, web standards, peer-reviewed research, and vendors' own pages for their pricing and features. Our editorial standards
- Google Search Console Help. “Manual actions report.”
- Google Search Central. “Creating helpful, reliable, people-first content.”
- Yoast. “Word count and SEO: how long should an article or page be?”
- Google. “New ways we’re tackling spammy, low-quality content on Search.”
- Google Search Central. “Optimizing your website for generative AI features on Google Search.”
- Google Search Console Help. “Page indexing report.”
- Google Search Console Help. “Performance report (Search results): Overview and basic setup.”
- Google Search Central. “Site moves and migrations.”
- Google Search Central. “Block Search indexing with noindex.”
- Google Search Console Help. “Search Console annotations.”
