Thin Content: Identifying and Fixing Low-Value Pages

Why thin content becomes a risk for B2B websites

Thin content is one of the oldest problems in SEO and still one of the most relevant. Since the Google Panda update in 2011, Google has systematically downgraded pages with little added value. The wave of AI-generated text gave the topic new urgency in 2024. Google confirmed that after the March 2024 core update, search results should show 45% less low-quality, unoriginal content, instead of the 40% originally expected (US/international).

For B2B companies in the DACH region, this is not an academic problem. Service pages, location pages and landing pages that consist only of boilerplate and repeated keywords lose visibility or never get indexed in the first place. At the same time, teams under time pressure quickly produce thin content when they use AI tools without editorial control. This article shows you how to identify, fix and systematically prevent thin content.

What thin content really is: added value, not word count

The most common misconception is that thin content is simply content that is too short. That is wrong. Google does not measure quality by word count. Google's John Mueller made clear that the word count of a page is not a quality factor and not a ranking factor (US/international). So a short page is not automatically thin. The reverse also holds: even a 2,000-word text can be thin content if it delivers nothing new.

The precise definition comes from practice. According to Semrush, thin content pages are web pages with little or no value that do not satisfy search intent and do not help users complete a task (US/international). Two criteria are decisive: does the page satisfy search intent, and does it offer independent added value compared with what is already on the web?

A real-world figure helps put the word count debate into perspective. According to the HTTP Archive Web Almanac 2024, the median was 400 words on desktop home pages and 333 words on inner desktop pages (US/international). This shows that typical good pages are often shorter than SEO myths suggest. Word count is a symptom, not a diagnostic criterion.

The most common types of thin content

Thin content follows recurring patterns. If you know them, you will find them faster in your own content inventory.

• Doorway and keyword-only pages: Pages created only to rank for specific keywords, such as "SEO agency Vienna", "SEO agency Graz", "SEO agency Linz" with identical text and only the city name swapped. They pass users along without real content.
• Thin affiliate pages: Content that only promotes third-party products without its own tests, data or context. According to Semrush, thin affiliate pages with little original content are among the classic types of thin content (US/international).
• Automatically or AI mass-generated pages: Texts produced in large numbers without editorial review. They are often grammatically correct but interchangeable in substance.
• Tag, archive and pagination pages: Automatically generated overview pages in content management systems that carry hardly any content of their own and bloat the index.
• Boilerplate and duplicate content: Standard texts repeated almost identically across many pages, such as generic industry descriptions or copied manufacturer texts.

Identifying thin content: the audit workflow

A structured audit combines performance data with crawl analyses. The scale of the problem is often underestimated. According to an Ahrefs study, 96.55% of all pages get no organic traffic from Google, and another 1.94% get only one to ten visits per month (US/international, index of around 14 billion pages). A large share of these pages is thin content or misses search intent.

Here is how you go about it:

• Analyze Google Search Console: Filter pages by impressions, clicks and average position. Pages with many impressions but almost zero clicks point to missed search intent. Pages without a single impression are candidates for pruning.
• Use crawl tools: Screaming Frog, Semrush or Ahrefs identify pages with low word count, missing metadata, thin internal linking and duplicate content clusters. Important: low word count is only a filter, not a verdict.
• Check engagement signals: Dwell time, bounce rate and scroll depth from your web analytics show whether users find the content helpful.
• Analyze index bloat: Compare indexed pages (site: query, GSC coverage report) with the number of strategically relevant pages. A large gap points to tag, filter and archive pages that are indexed unnecessarily.

AI content: what Google rewards and what it penalizes

AI writing has long been everyday practice in the DACH region. According to Statistik Austria, 20% of Austrian companies with ten or more employees already used AI technologies in 2024, almost twice as many as in 2023 at just under 11% (Austria). Of the companies using AI, 65% use text recognition and processing and 41% use language generation (Austria). AI content is therefore a mass reality, not the exception.

Google does not rate AI content as negative across the board. What matters is the purpose. With the March 2024 core update, Google introduced the scaled content abuse spam policy: it applies to content created in large numbers primarily to manipulate rankings, regardless of whether it comes from automation, human work or a combination that includes generative AI (US/international). The tool is not the problem. The intent and the missing original contribution are.

An analysis by Originality.ai shows how real the risk is. Of around 79,000 sites examined, about 1.9% received a manual action, and 100% of these affected sites showed signs of AI content use (US/international). That does not mean AI use alone gets penalized. It means that uncontrolled, volume-focused AI content is a clear warning sign. Human-in-the-loop editing is therefore mandatory, not optional: check facts, add your own contribution, verify sources.

Information gain and E-E-A-T as the countermeasure

The most effective remedy for thin content is measurable added value. A patent shows how Google could evaluate it. According to Search Engine Land, a Google patent on an "information gain score" was granted in June 2022 (US/international). According to the patent specification, the information gain score evaluates how much additional information a document contains compared with the documents the user has already seen. Whether and how Google uses it in ranking is not confirmed, because a granted patent does not prove use. The principle does match Google's guidance on helpful content. There, Google asks whether content provides original information, research or analysis and offers substantial value compared with other pages in the search results. Interchangeable AI content and content farms fail precisely on this point.

In practice, information gain means: bring in something that does not exist yet. That includes original data from your own projects, first-hand information from your own industry, attributable expert quotes and an independent perspective. E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) is the structural frame around this: demonstrable experience and expertise make the difference between an interchangeable summary and a reference source.

Fixing thin content: improve, consolidate or delete

There are three options for every thin page. A simple decision tree helps.

• Improve and enrich: If the page covers a strategically relevant topic with real search volume but is too thin, add your own contribution. Add original data, examples, FAQs and expertise. Simply padding the page with more text does not help, as John Mueller stresses: blindly adding more and more text to a page doesn't make it better (US/international).
• Consolidate: If several thin pages serve the same topic or the same search intent, merge them into one strong page. Redirect the weak URLs to the main page with a 301 redirect or set a canonical tag. This also eliminates keyword cannibalization.
• Delete or noindex: Pages without added value and without a strategic purpose are removed (with a 301 to a sensible alternative) or taken out of the index with noindex, for example tag and filter pages.

An Ahrefs case study on Belkins shows that content pruning works. After removing around 400 low-quality pages, almost two thirds of the entire website, organic traffic rose from 3,000 to almost 10,000 visits per month (US/international). Fewer but better pages are often the stronger strategy.

Prevention at site level

You can prevent thin content structurally. Three levers are central.

• Topic clusters and pillar pages: Structure content around central pillar pages with supporting cluster articles. This prevents isolated, thin standalone pages and concentrates topical authority.
• Avoid keyword cannibalization: Only one target page per search intent. Maintain a keyword map before new content gets created.
• Editorial standards and briefings: Define minimum requirements for each content type: which original contribution, which sources and which information gain are mandatory. For AI-assisted production, a mandatory editing step belongs in the briefing.

Zero-click and AI Overviews: standard content disappears

The SERP has changed. Interchangeable standard content is summarized directly in AI Overviews and zero-click answers, and the click to the source never happens. According to an Ahrefs study, the presence of an AI Overview correlates with a 34.5% lower average click-through rate for the top-ranking page (US/international, 300,000 keywords analyzed). The generic content that AI can summarize effortlessly is exactly what loses the most.

The consequence: added value has to be visible and impossible to reproduce. Original data, concrete case figures, your own methodology and expert assessments are harder to replace in an AI answer. If you only repeat the obvious, you supply the material for the summary without getting the click.

DACH specifics: EU AI Act and a legally compliant AI strategy

For companies in Austria and Germany, there is also a legal dimension, and for Swiss companies when the output of their AI systems is used in the EU. The deployer obligations under Article 50 of the EU AI Act have applied since August 2, 2026. As a deployer, you have to disclose deepfakes and label AI-generated text that you publish to inform the public on matters of public interest. For such text, the obligation does not apply if a human has reviewed it or exercised editorial control and a natural or legal person holds editorial responsibility. Machine-readable marking of AI-generated content, by contrast, is the job of the providers of the AI systems. A clean AI content strategy therefore combines two goals: SEO quality (original contribution, information gain) and legal compliance (transparency, labeling where required).

In practice, this means: document your editorial process, keep verifiable records of human-in-the-loop steps and check for each content type whether labeling applies. This discipline protects you from thin content and from regulatory risk at the same time.

B2B practice: service, location and landing pages

In a B2B context, thin content typically arises in three places: service pages with generic descriptions, location pages built on the doorway pattern and landing pages without real content. Programmatic SEO, meaning the template-based creation of many similar pages, is not thin content per se. It stays legitimate when every generated page carries its own useful data, such as real local information, specific use cases or structured facts. Without this data core, programmatic SEO turns into a thin content factory.

Metrics for measuring content quality

Content quality needs measurable KPIs. Track them per page and in aggregate:

• Clicks per impression (CTR from GSC): A low CTR with high impressions signals missed search intent.
• Indexing rate: The share of submitted pages that actually get indexed. A falling rate points to quality problems or index bloat.
• Share of pages without traffic: How many URLs get zero organic visits. This list is your pruning pipeline.
• Engagement: Dwell time and scroll depth as a proxy for perceived added value.
• Rankings per search intent: A clear mapping prevents several pages from competing for the same keyword.

Further reading

Thin content is not a word count problem. It is an added-value problem. If you take information gain seriously, use AI only with editorial control and prune consistently, you win in the era of AI Overviews. Go deeper with the related articles on information gain, E-E-A-T, topic clusters and pillar pages, and content strategy under the EU AI Act.