I was reading X the other day and came across a few tweets from a pair of prominent SEO professionals. I found the statements interesting, not because they were flat-out wrong, that would be too easy, but because they highlighted one of the recurring communication problems in SEO.
It is often not the sentence itself that causes the trouble. It is what gets lost between the lines. The nuance, the awkward middle, the bit that refuses to fit neatly into a screenshot, a course module, or some dopamine-friendly thread designed to make people nod along and feel clever.
The broad idea being argued was this: HCU did not touch 'thin content', because most sites within the HCU strike zone had plenty of word count. Thin content, in this telling, is little more than copywriter-industry FUD, a marketing campaign built to frighten site owners into buying more words. Google, we are told, is not a content appreciation engine. It uses maths. It cannot know whether content is good or bad, because content quality is subjective. And because of that, nothing has really replaced PageRank, except in the minds of people who do not understand it, or do not like it.
Now, there is a kernel of something useful in there. Quite a big kernel, actually. But like most strong statements in SEO, it starts to wobble the moment you stop treating it as a pub argument and start treating it as a thing that needs to survive contact with primary sources.
The problem is not that the statement is useless. The problem is that it is too clean.
And in SEO, clean explanations are often where understanding goes to die.
Thin, is not Skinny
The useful part of the argument is simple enough. Thin content is not the same thing as short content. Word count is a poor proxy for usefulness, quality, depth, authority, experience, or any of the other words that tend to get thrown around when SEOs are trying to describe why a page did, or did not, survive a Google update.
Google representatives have said this repeatedly. John Mueller has said that word count is not a sign that a page is thin content, and that site owners should not use word count to decide what is helpful or what is fluff.1 He has also said that word count is not a quality factor.2 In 2025, he went further again, saying that word count itself makes very little sense as a diagnostic measure.3
That should kill a lot of lazy SEO advice stone dead. It does not, of course, because there is always another spreadsheet, another content brief, another 'minimum 1,500 words' instruction that has been passed around like some sacred relic from the days when people still thought keyword density was a personality.
But here is the rub. Saying that 'thin content' should not be defined by word count is not the same as saying that thin content has no foundation. Google has used the phrase 'Thin content with little or no added value' in its own Search Console manual-action documentation.4 Google also described the Helpful Content Update as a system that identified content with little value, low-added value, or content that was otherwise not particularly helpful to searchers.5
So the sharper, more accurate position is this: the HCU did not punish pages for being short. It was not a word-count update. But it was very much connected to content, if by content we mean usefulness, originality, added value, satisfaction, evidence, experience, and whether the page deserved to exist beyond the fact that a keyword tool said it could bring traffic.
| The statement | The more useful reading |
|---|---|
| HCU did not touch thin content because affected sites had lots of words. | HCU did not target short content, but long content can still be thin if it adds little value. |
| Thin content is copywriter FUD. | Word-count panic is FUD. Low-added-value content is a real Google concept. |
| HCU had nothing to do with content. | HCU was not about literary quality, but it was clearly about helpfulness and usefulness as inferred by systems. |
| Google cannot know subjective content quality. | Google cannot know universal human taste, but it can estimate relevance, reliability and likely satisfaction through signals. |
| Google uses maths, so it knows nothing about content. | Google uses maths to model things that humans care about, imperfectly, at scale. |
| Nothing replaced PageRank. | PageRank still matters, but it now lives inside a much larger ranking ecosystem. |
That is where the understanding lives. Not in the slogan. In the middle bit.
The core distinction: short content is not the same as thin content
This is the part the tweets get right, or at least close enough to right that we should not throw the whole thing out.
A short page can be the best result on the internet for a query. If someone searches for the opening hours of a business, a VAT threshold, a CSS property, a recipe temperature, a simple definition, or a yes-or-no answer, they do not need the bones of a dissertation strapped to the page just so an SEO tool can flash green.
Likewise, a long page can be absolute fluff. It can be stuffed with introductory throat-clearing, rewritten competitor headings, stock definitions, AI-generated filler, affiliate padding, FAQs nobody asked, and that peculiar type of content that seems to exist purely because someone exported a keyword cluster and mistook it for a publishing strategy.
John Mueller’s 2022 wording is worth sitting with:
“Word count is not a sign that a page is thin content. You're the expert on your site's topic... you can make a qualified call on what's helpful for users, and what's fluff. Don't use word count.”1
That is not a complicated statement. Yet the industry has spent years finding ways to avoid understanding it, because word count is seductively easy. It gives managers a number. It gives copywriters a scope. It gives SEOs a deliverable. It gives everyone a little comfort blanket.
But content does not become useful because it got longer. A bad answer with more paragraphs is still a bad answer. A derivative article with a few extra subheadings is still derivative. A page built from 'People Also Ask' boxes is not automatically insight, it is often just archaeology performed on someone else’s ranking success.
The phrase 'thin content' only becomes useful when it is separated from length. Thin means little added value. Thin means no meaningful experience. Thin means no evidence. Thin means a page that says the same thing as everyone else, but with a different logo at the top. Thin means a site that has learned the shape of helpfulness without carrying the weight of it.
What Google actually said HCU was about
The original Helpful Content Update announcement was not subtle. Google called it 'More content by people, for people in Search'.5 That alone makes the claim that HCU had 'nothing to do with content' difficult to hold without doing a fair bit of verbal gymnastics.
Google described the system as automated and machine-learning based. It was not a manual action. It was not a spam action. It was one signal among many used to rank content.5
“This classifier process is entirely automated, using a machine-learning model. It is not a manual action nor a spam action. Instead, it's just a new signal and one of many signals Google evaluates to rank content.”5
That matters, because it tells us two things at the same time. First, HCU was connected to content helpfulness. Second, it was not a human editor sitting there with a red pen, deciding whether your article had enough soul.
This is where SEO conversations often split into two camps, both of them slightly allergic to nuance. One camp talks as if Google can read content like a wise old librarian, quietly appreciating insight, craft and lived experience. The other camp talks as if Google is just PageRank in a trench coat, blind to everything except links and mathematical structure.
Both views are too tidy.
The more boring, and therefore more likely, answer is that Google builds systems that infer. It does not need to appreciate an article in the human sense. It needs to decide, at scale and in a fraction of a second, which document is more likely to satisfy a query. That is not taste. It is modelling.
So, was HCU about content? Yes. Was it about content in the way many copywriters and consultants sell content quality? No.
That difference is the whole game.
What affected sites seemed to have in common
The September 2023 HCU was especially brutal because many site owners looked at their pages and could not see the crime. They had articles. They had headings. They had word count. They had internal links. They had schema. They had all the little SEO artefacts that make a page look respectable under fluorescent agency lighting.
And still, traffic fell off the table.
That is probably why the 'HCU did not touch thin content' argument has some emotional power. Many affected sites were not thin in the old, childish sense of having 120 words and a dream. They often had thousands of pages, serious topical coverage, long guides, comparison content, affiliate reviews, how-to material and substantial publishing histories.
But again, having more content does not mean having more value.
| Pattern seen across affected sites | Why it matters |
|---|---|
| Programmatic or AI-generated content at scale | High publishing volume can create the appearance of coverage while producing many pages with little originality or human usefulness. Glenn Gabe reported AI-generated and programmatic content patterns among affected sites.9 |
| SEO-driven topical sprawl | A site can cover every imaginable variation of a topic and still add very little beyond what already exists in the SERP.9 |
| Excessive ads and poor user experience | Disruptive ads, popups, auto-playing video and buried main content were repeatedly noted in industry analysis around HCU-hit sites.9 10 |
| Affiliate or review content without real evidence | Long reviews are still thin if they contain no proof of testing, no original analysis and no clear experience with the product.10 |
| Lack of experience or information gain | A well-written guide can still struggle if it is one more interchangeable answer in a SERP already full of interchangeable answers.11 |
This is the bit that tends to annoy people, because it removes the comforting simplicity of the diagnosis. It means the answer is not 'write more'. It is also not 'delete everything under 700 words'. Both are lazy.
The actual question is more uncomfortable. Why should this page exist? What does it know that the other pages do not know? What has been experienced, tested, observed, bought, broken, repaired, learned, regretted, or earned? What is the information gain? What is the human residue on the page?
That phrase, 'human residue', is probably not in any Google documentation, but it should be in more SEO conversations. Because a lot of the web now has the opposite problem. It has the shape of expertise, but none of the scar tissue.
Did HCU have 'nothing to do with content'?
No. That is too far.
Google’s own helpful content guidance asks whether content provides original information, reporting, research or analysis, whether it gives a substantial or complete description of the topic, whether it provides insight beyond the obvious, whether it avoids simply copying or rewriting sources, and whether readers leave feeling satisfied.12
Those are content questions. They are not word-count questions. They are not purely PageRank questions. They are not a checklist for literary beauty either, but they are absolutely questions about whether the thing on the page deserves to be ranked.
Where the confusion creeps in is that HCU did not appear to behave like a neat page-by-page editorial judgement. Google described the original system as using a site-wide signal.5 Later, Google said helpfulness had been folded into core ranking systems using a variety of signals and approaches, with systems primarily designed to work at page level while also using site-wide signals.8
So the truth is less convenient than the slogan. HCU was not a simple content audit. It was not a copywriting score. It was not a 'your article is bad' sticker applied by some invisible examiner. But it was about patterns of helpfulness and unhelpfulness across content ecosystems.
That word matters, ecosystem.
A website is not just a pile of URLs. It is a publishing culture. It has incentives. It has habits. It has blind spots. It has commercial pressures. It has a relationship with the reader, even when that relationship is never consciously acknowledged. HCU seemed to punish, or at least expose, sites where the publishing culture had drifted too far towards search-engine-first production.
Not because every sentence was bad. Because the system, taken as a whole, looked like it was built for extraction rather than usefulness.
Can Google 'know' content quality?
This is where the original argument becomes interesting again.
Can Google know whether content is 'good' for three billion people? No, not in the romantic, human, subjective sense. Google is not sitting in Mountain View having a quiet emotional response to your paragraph structure. It is not moved by your turn of phrase. It does not admire your analogy. It does not feel the small spark of recognition that happens when a writer actually says something true.
Google itself says its systems look for quantifiable signals and are not designed to analyse subjective concepts such as viewpoint or political leaning.6 That is important. It prevents us from drifting into the fairy tale that Google is a content appreciation engine.
But the opposite fairy tale is just as weak. The fact that Google cannot know quality as a human truth does not mean it knows nothing. Google says its ranking systems look at meaning, relevance, quality, usability and context.6 It also says its systems seek signals that help determine whether content demonstrates expertise, authoritativeness and trustworthiness, including whether other prominent websites link or refer to the content.6
Google also uses Search Quality Raters to evaluate whether results meet search needs and quality standards. Those ratings do not directly change rankings, but they are used to benchmark and improve Google’s systems.7
So, no, Google does not have universal taste. It has proxies. It has signals. It has models. It has testing. It has raters. It has endless mathematical machinery trying to approximate what humans might find useful, credible, satisfying or safe enough to show near the top of a results page.
That is not magic. It is not wisdom. It is not Deep Thought handing us 42 from the clouds.
It is a machine making probabilistic judgements inside a commercial search ecosystem.
And that is enough to rank or tank a site.
PageRank still matters, but it is not the whole story
Now to PageRank, the old ghost in the machine.
The statement that nothing replaced PageRank has a certain appeal, especially to those of us who remember when links were the cleanest way to cut through the culture soup of SEO nonsense. And to be fair, PageRank has not disappeared. Google’s own ranking systems guide says PageRank was one of its core ranking systems when Google first launched, that it has evolved a lot since then, and that it continues to be part of Google’s core ranking systems.8
“Among these is PageRank, one of our core ranking systems used when Google first launched... How PageRank works has evolved a lot since then, and it continues to be part of our core ranking systems.”8
That should stop anyone claiming links no longer matter. They do. Authority still matters. The structure of the web still matters. Who points at you, who cites you, who trusts you, who repeats you, who ignores you, all of that still sits somewhere inside the great tangled mess.
But it is equally wrong to reduce modern Google to PageRank with a fresh coat of paint. On the same ranking systems page, Google lists BERT, neural matching, RankBrain, passage ranking, original content systems, reliable information systems, reviews systems, site diversity systems, spam systems and more.8
So PageRank is not dead. But neither is it the whole animal.
The better mental model is not replacement. It is layering. Google did not throw away links and become a content poet. It layered language understanding, entity understanding, quality systems, spam systems, freshness, locality, review quality, originality and user-satisfaction proxies on top of a web that is still deeply shaped by links.
Which is annoying, of course, because layered systems are harder to sell in a tweet.
Broad brushes lose fine detail
The strongest part of the original argument is its refusal to worship word count. That instinct is right. The SEO industry has spent far too long confusing volume with value, and too many copywriters have benefited from the fear that every short page is somehow a liability.
But the argument overcorrects.
Thin content is not a myth. Thin content as 'not enough words' is the myth. HCU was not a word-count update, but it was absolutely connected to content helpfulness. Google does not appreciate content like a human reader, but it does use automated systems to estimate whether content is useful, relevant, reliable and satisfying. PageRank still matters, but it is not the only thing sitting behind modern ranking.
The more honest version is this:
HCU was not a penalty for short pages, and 'thin content' should never be diagnosed by word count. The real risk is low-added-value content at scale, pages that are derivative, over-templated, search-engine-first, experience-poor or unhelpful relative to what already exists. Google does not appreciate content like a human critic, but it does use automated systems and signals to estimate which content is likely to satisfy users.
That is less punchy than the original claim. It will not travel as well. It does not have the same satisfying edge.
But it is closer to the truth.
And in SEO, that is usually where the hard work begins.
