Thin Content SEO: What Counts as Thin and How to Fix It

"Thin content" is Google's term for pages that don't offer enough original value to earn a place in search results — it's not a measurement of length, it's a measurement of substance. Site owners often treat it as a word-count problem and chase a number, then miss the real issue: a 2,000-word page padded with generic advice and a tight 200-word page that answers one specific question precisely can trade places in Google's judgment, with the shorter one winning. Google's spam policies for web search name several thin-content patterns directly — thin affiliate pages, auto-generated content produced without human oversight, and doorway pages that repeat a template across locations, products, or tags — and its broader quality systems apply similar judgment more loosely, without a formal notice, by simply not surfacing a page.

This article works through what each pattern actually looks like on a real site, how to find it once you have more than a handful of pages, and which fix — expand, consolidate, or noindex — resolves it instead of just disguising it. If you already know a page has to go and the only open question is whether to delete it outright or merge it into another URL, that's a pruning decision, not a diagnosis; this piece is about recognizing thin content in the first place and matching it to the right repair.

What Google Means by "Thin"

Thin content adds nothing beyond what a searcher can already get from a page Google has already indexed for that query. It's a relative judgment, not an absolute one: the same 400 words can be thin on a competitive query where twenty other pages say the identical thing, and sufficient on a narrow question nobody else has answered directly.

None of this is defined by length. A page can be long and still be thin — padded with restated points, filler transitions, and a definition paragraph that could apply to a thousand other pages — and a page can be short and not be thin at all, if it answers the one question it exists for and stops. Treating "thin" and "short" as synonyms is the mistake to avoid early, because a length-driven fix doesn't touch the actual problem and can make some patterns worse instead of better. Google's spam policy documentation instead describes a handful of concrete patterns, each with a different fix:

• Auto-generated content produced at scale without a human adding value or checking accuracy
• Thin affiliate pages that repeat a manufacturer's description and attach a buy link, with no independent evaluation
• Doorway pages built to rank for many near-identical keyword, location, or product variants that all funnel to the same underlying offer
• Scraped or lightly reworded content, and pages that exist mainly to hold ads with content incidental to that purpose

Thin Affiliate Content

The clearest version of this pattern is a product review or "best X" page that reads like the manufacturer's own spec sheet with a buy button attached: the paragraphs restate features already listed on the retailer's page, there's no evidence anyone tested the product, and the only original sentence is the affiliate disclosure that advertising rules typically require. Google's guidance singles this out because it's common at scale — an affiliate site can produce hundreds of these pages by lightly rewording manufacturer copy, none adding anything a searcher couldn't get by going straight to the retailer.

A common near-miss is running the page through a paraphrasing tool to change wording without changing the information content — it reads differently but still says nothing new, because the issue was never the phrasing. What actually resolves a thin affiliate page is genuinely additive, not cosmetic:

• Independent evaluation — how the product performed in an actual use case, not a restatement of the spec sheet
• Comparative reasoning against named alternatives, including where the product loses, not just where it wins
• Specific detail that couldn't have been copied from the listing: setup friction, a defect noticed only after a week of use, a use case the manufacturer doesn't mention
• Pricing context that's accurate and current, never an invented range presented as market data
• When a site has many single-product pages this thin and none individually justifies a standalone URL, consolidating them into a comparison or buying-guide page that can support real evaluative content across several products at once

Auto-Generated and Templated Pages

This pattern shows up whenever a page is produced by filling a fixed template with values pulled from a database — a paragraph structure that stays constant while a city name, product category, or search term gets substituted in. The signature: two pages from the family read identically once you strip out the substituted term. A directory site generating a page for every "service" and "city" combination in its database, using the same five paragraphs each time, is producing this pattern regardless of whether a person or a script wrote the original template.

It's worth being precise about Google's actual concern here, because it's not the production method. The policy language targets scaled abuse — using automation, with or without an AI model involved, to manufacture ranking-manipulative content at a volume no editorial process is checking. A page drafted with AI assistance but then fact-checked and given real specifics by someone who knows the subject is not what the policy describes. A page produced entirely by a human, following a rigid template with no per-page substance, absolutely is. The test is whether real, page-specific value was added, not which tool typed the first draft.

The fix depends on whether the substituted variable actually supports unique content. If your data can genuinely differentiate the pages — different local regulations, different real pricing, different use cases — expand each with that material and the URLs earn their place. If it can't (most "[City] Plumber" pages fall here, since the service doesn't meaningfully differ by dispatch city), the honest fix is reducing the page count to what the data actually supports — usually one strong resource with a filter or directory instead of hundreds of near-duplicate URLs competing for the same intent.

Boilerplate Location and Tag Pages

A close cousin of the templated-page pattern: service businesses generating a page per city served, and e-commerce or blog sites generating an archive page per tag or filter combination, where the only thing that changes between pages is a proper noun dropped into an otherwise fixed paragraph. The diagnostic is fast: open two side by side. If the only distinguishing text is the city name, business name, or tag label, the page is boilerplate no matter how polished the paragraph reads.

Two situations call for two different fixes:

• If genuine local or category-specific substance is available — a documented service area, a testimonial tied to that location, a photo of a completed job there — expand the page with it, and it becomes a legitimate standalone URL
• If no such substance exists or won't scale to every page, consolidate into a single "Service Areas" or category page with a clear list or map, and 301-redirect the retired URLs into it
• For tag pages specifically, a tag with only one or two items rarely deserves its own indexed archive; noindexing sparse ones, or merging overlapping tags into a smaller set with enough content behind each, removes the boilerplate without losing the organizational value tags provide visitors

How to Find Thin Content on Your Site

Diagnosing this at the scale of a real site — hundreds or thousands of URLs — needs a process, not a page-by-page read-through:

• Crawl the site with a tool that reports word count per URL and treat that count only as a screening signal to sort pages for review, never as the verdict itself
• Cluster pages by title or H1 pattern — a regular expression matching "[City] [Service]" or "[Brand] [Category]" surfaces entire templated families in one pass, far faster than reviewing pages individually
• Pull Search Console's Page indexing report and filter for "Crawled – currently not indexed" and "Discovered – currently not indexed," then narrow by URL pattern; the first means Google crawled a page and chose not to index it, the second that Google has not yet prioritized crawling it, and a cluster sharing a pattern is the doorway or templated-page signature
• For any suspected family, strip the substituted variable mentally, or diff two URLs directly, and check whether the remaining text is distinguishable — if it isn't, that's confirmation, not just suspicion
• Cross-check with analytics for near-zero organic sessions over a long window; this corroborates a diagnosis but shouldn't stand in for one, since a page can get little traffic for reasons unrelated to content quality, such as being new

The Fix: Expand, Consolidate, or Noindex

Once a page is confirmed thin, the decision that matters is which of three repairs resolves the specific pattern it fell into — the wrong one either wastes the effort or leaves the underlying problem in place. The three are not interchangeable: noindexing means a page should keep existing but stop competing for search visibility, consolidating means its content should live somewhere else, and deleting means neither the page nor its content should continue to exist. Which applies to a page already flagged as thin turns on things this diagnosis doesn't answer by itself, like whether other pages link to it and whether any equity there is worth redirecting.

• Expand when the topic genuinely deserves its own URL and real, specific content exists to add — original analysis, a documented use case, direct answers to sub-questions a searcher would actually have. Right call for a thin affiliate page where independent testing is possible, or a location page where real local detail exists.
• Consolidate when several thin pages cover overlapping ground and none individually earns a place in the index — merge the useful pieces into one comprehensive page, and 301-redirect the retired URLs so you don't strand whatever inbound links they'd accumulated.
• Noindex when a page needs to keep existing for actual users — a location's contact details a current customer needs, an old product page still linked from a receipt — but has no independent case for ranking. A noindex, follow directive keeps the page live and reachable directly while keeping it out of search results.

A Worked Example

Consider a plumbing company with forty pages, one per city it serves, each roughly 300 words and built from the same five paragraphs with the city name and a local phone number swapped in. A word-count crawl doesn't flag anything on its own — 300 words is a plausible length for a legitimate local page. What flags it is the H1 pattern: every one matches "[City Name] Plumbing Services," and a Search Console Page indexing export shows thirty-three of the forty sitting in "Crawled – currently not indexed." Opening two side by side confirms it: past the city name and phone number, every sentence is identical, including a claim of coverage of "the entire metro area" on pages for cities forty miles apart.

The fix isn't uniform, because the underlying situation isn't. The five cities generating the most actual business — verifiable through call volume, not invented figures — get expanded with real material: a specific service-area description, a testimonial from an actual customer there, a note about a local permitting quirk the company has handled. Those five earn standalone pages. The remaining thirty-five, where no city-specific detail exists, get consolidated into a single "Areas We Serve" page listing all forty cities with a simple map, and the retired URLs get 301-redirected into it rather than deleted outright, preserving whatever minor link equity a few had picked up.

Mistakes That Make Thin Content Worse

A few patterns show up repeatedly when sites try to fix thin content and end up compounding the problem instead:

• Running pages through an AI rewriting or "expansion" tool purely to hit a word count, producing padded prose that restates the same information without adding anything new — this is cosmetic, the page is still thin, and doing it across many pages at once sits closer to the scaled-abuse pattern than a fix for it
• Noindexing every flagged page without checking whether it independently drives value — a booking or contact page that ranks poorly but converts the direct traffic it gets is not a candidate for the same treatment as a page with zero function beyond ranking
• Consolidating pages without setting up 301 redirects from the retired URLs, orphaning any existing backlinks and discarding the small amount of authority they'd accumulated instead of transferring it
• Fixing the content decision but not the surrounding signals — leaving a noindexed or redirected page in the XML sitemap, or internal links pointing at a URL since merged elsewhere, so crawlers keep rediscovering the state you were trying to change