An SEO myth is a ranking rule that circulates as fact without anyone ever having traced it back to its source. It survives because it is convenient: it gives a simple explanation for a system that is not simple, and repeating it costs less than checking it.
Checking it usually takes ten minutes. Almost every belief collected here is contradicted by a public Google document, with a date and an address. Where the documentation is not enough, I measured the UK SERP directly on 26 September 2026.
Why SEO myths survive everything
I went through the pages that currently rank for "seo myths" on Google UK, leaving out a LinkedIn post and a Reddit thread. Of the four agency articles I could read in full, only one links to a Google document. The others quote Google from memory, or not at all.
The dates explain the rest. The first article on the page is a news item from 5 December 2016 that summarises a list published by another SEO blog. Of the others that show a date, the most recent is from 1 August 2026, and the list also includes 1 April 2026, 9 July 2026 and 27 August 2025. Above all of them sits an AI Overview.
The result is a closed loop. One page says the meta keywords tag is "completely ignored by Google" without saying where Google wrote that, the next repeats it, the third inherits it. None opens the 2009 post in which Google put it in writing. When the source is missing, repetition takes its place, and that is how an invented rule becomes something "everyone knows".
Eleven myths and the document that disproves them
This table is the short version. Below, each row gets its own explanation and its own link.
| The myth | What the source says | Where it is written |
|---|---|---|
| Domain authority decides who comes first | On the SERP we measured, rank correlation between position and domain score: 0.06, i.e. none | Visilay measurement on DataForSEO data, 26 September 2026 |
| Meta keywords still matter | "Google does not use the keywords meta tag in web ranking" | Google Search Central, September 2009 |
| There is an optimal keyword density | No threshold. Only the definition of keyword stuffing | Google Search spam policies |
| Google penalises duplicate content | Duplicates are a problem of choosing the canonical URL, not a penalty | Canonicalisation documentation |
| Structured data pushes you up the rankings | "Using structured data enables a feature to be present, it does not guarantee that it will be present" | Structured data general guidelines |
| Under a thousand words you won't rank | No minimum word count appears in the documentation | Google Search Central (absence checked) |
| Core Web Vitals decide rankings | "There is no single signal" and relevance comes before page experience | Page experience documentation |
| The meta description is a ranking factor | "Snippets are primarily created from the page content itself" | Snippet documentation |
| Google penalises AI-written text | "Our focus on the quality of content, rather than how content is produced" | Google Search Central Blog, February 2023 |
| AI Overviews have killed SEO | Clicks on organic results drop from 15% to 8% of visits when an AI summary appears (US), and in Italy 40% of cited links come from outside the top 10 | Pew Research Center, July 2025; SEOZoom study of over 30 million keywords, December 2025 |
| If you're in the top 10, AI will cite you | 175,000 domains cited out of 168 million monitored in Italy, 0.1% | SEOZoom study, December 2025 |
Myth 1: domain authority decides who comes first
Moz's Domain Authority and Ahrefs' Domain Rating are metrics from two private companies that estimate the strength of a link profile. Google does not calculate them, does not receive them and does not mention them anywhere in its documentation. They are useful for comparing two sites at a glance, not for predicting an order on the SERP.
The quickest way to see this is to take a SERP and compare the order with the scores. I did it for the query "seo myths", Google UK, desktop, on 26 September 2026, crossing the organic results with DataForSEO data on each domain's link profile.
| Position | Domain | Domain score (0-100) | Referring domains to the site | Referring domains to the page |
|---|---|---|---|---|
| 2 | cosmic.org.uk | 62 | 706 | 0 |
| 4 | verto.co.uk | 29 | 125 | 0 |
| 5 | vidigital.co.uk | 48 | 76 | 0 |
| 6 | seekmarketingpartners.com | 43 | 544 | 0 |
| 7 | pearldigitalmarketing.com | not returned | 74 | 0 |
| 8 | whitespark.ca | 54 | 6,381 | 8 |
| 9 | coalitiontechnologies.com | 54 | 3,044 | 0 |
The rank correlation between position and domain score is 0.06, which is to say none. Between position and referring domains it is 0.36, and it points the wrong way: the more linked-to sites sit lower down. The two strongest link profiles on the page, with 6,381 and 3,044 referring domains, are in eighth and ninth place. The site with the lowest score, 29, is fourth, ahead of four sites with higher scores.
The figure I find most interesting is the last column. Six pages out of seven do not have a single backlink pointing to the specific URL. They rank with zero links to the page. Links matter, but they matter for relevance and at the level of the individual page, not as an overall score to be nudged up a few points a year.
Myth 2: meta keywords are still good for something
In September 2009 Google published a post titled Google does not use the keywords meta tag in web ranking. The title is the whole statement.
Seventeen years later the field still appears in plenty of templates and plugins, often carefully filled in by whoever runs the site. It does no harm, but it is time spent on a tag that no search engine has read since before the iPad existed. If you are tidying up the tags on a page, the attention belongs on the meta title, which does still count.
Myth 3: there is a keyword density to hit
The famous 2-3% does not appear in any Google document. What does appear, in the spam policies, is the definition of the opposite behaviour: "Keyword stuffing refers to the practice of filling a web page with keywords or numbers in an attempt to manipulate rankings in Google Search results".
The examples Google lists are lists of phone numbers, blocks of cities and regions put there to rank, and phrases repeated "so often that it sounds unnatural". No threshold, because there is no counter: there is a judgement on whether the text sounds artificial. I have written about it in detail on the pages about keyword density and keyword stuffing.
Myth 4: Google penalises duplicate content
The spam policies list sixteen categories of behaviour: cloaking, doorways, scraping, link spam, hidden text, expired domain abuse, keyword stuffing and others. "Duplicate content" is not among them. Scraping is, but that is something else: copying from other people's sites.
In the canonicalisation documentation Google treats duplicates as what they are, a problem of choice: "Google will identify which version of the URL is objectively the best version to show to users in Search". The real risk is not a penalty, it is Google choosing a different version from the one you would have chosen. You manage it with the canonical tag, and the full set of cases is in the guide to duplicate content.
Myth 5: structured data pushes you up the rankings
Google's guidelines on structured data are clear on two points. The first: "Using structured data enables a feature to be present, it does not guarantee that it will be present". The second concerns manual actions for non-compliant markup, which remove eligibility for rich results but, in Google's words, "doesn't affect how the page ranks in Google web search".
In plain terms: schema markup changes how the result looks, not how high it sits in the list. That does not make it useless, because a result showing stars, a price or questions collects more clicks from the same position. It is CTR work dressed up as ranking work.
Myth 6: under a thousand words you won't rank
There is no minimum word count in Google Search Central's documentation. This is not a matter of interpretation: the threshold simply does not exist, on any page.
The SERP I analysed confirms it on its own. The page in fourth position is about 650 words long and gets through five myths; the one in sixth is roughly twice that. The right length is whatever it takes to answer, and you work it out by looking at what the pages already ahead of you cover, not from a threshold. The full reasoning is on the page about the ideal length of a blog post.
Myth 7: Core Web Vitals decide rankings
The page experience documentation contains three sentences worth reading before you sign off a quote for performance optimisation.
- "There is no single signal": there is no "page experience" factor with its own weight.
- "Google Search always seeks to show the most relevant content, even if the page experience is sub-par": relevance comes first.
- "Getting good results in reports like Search Console's Core Web Vitals report or third-party tools doesn't guarantee that your pages will rank at the top".
A slow site still needs fixing, because people leave. But going from a score of 72 to 95 on a page that answers the question badly moves nothing. Page speed is a condition, not a lever.
Myth 8: the meta description is a ranking factor
Google explains in its snippet documentation that "snippets are primarily created from the page content itself" and that the meta description comes into play only "sometimes", when it describes the page better than an extract from the text would.
You can see it with the naked eye. On the 26 September SERP, several snippets are not descriptive sentences at all but lists lifted from the body of the page: "Myth #1: SEO is something that needs to be set up once · Myth #2: ...". The meta description remains work on the click, and it should be measured as such: in Search Console, looking at CTR before and after, not at positions.
Myth 9: Google penalises AI-written text
It is the youngest myth and already the most widespread. Google answered it in February 2023, in a post signed by Danny Sullivan and Chris Nelson of the Search Quality team: "Our focus on the quality of content, rather than how content is produced, is a useful guide that has helped us deliver reliable, high quality results to users for years".
The line between legitimate use and abuse is written in the helpful content guide: "If you use automation, including AI-generation, to produce content for the primary purpose of manipulating search rankings, that's a violation of our spam policies". The category is called scaled content abuse, and the spam policies describe it as using generative tools "to generate many pages without adding value for users".
So the dividing line is not the tool, it is volume without added value. Fifty pages churned out in an afternoon to cover fifty keyword variants are exactly the behaviour described. A text written with AI, checked, supplemented with data the model could not have known and signed by someone, is not. I have collected the available research on how AI-assisted text actually ranks on the page about AI content and SEO.
Myth 10: AI Overviews have killed SEO
AI Overviews reached the UK in August 2024, after launching in the US in May of that year. On the query analysed above, one sits at the top of the page. The only large public measurement of coverage I know of comes from Italy: according to the SEOZoom study published in December 2025, built on more than 30 million Italian keywords, AI Overviews appear in 54.15% of searches there, up from under 1% on launch day in March 2025 and just over 11% in July 2025.
On traffic, the solid numbers come from the United States. The Pew Research Center followed 900 US adults and 68,879 real searches in March 2025: with an AI summary on the page, an organic result is clicked in 8% of visits, against 15% when there is no summary. Ahrefs, comparing 300,000 keywords between March 2024 and March 2025 on US desktop, measured a 34.5% lower CTR for position one when an AI Overview is present.
For the UK I have not found an equivalent public measurement of click behaviour. Anyone who tells you how much traffic you will lose in Britain is extrapolating from an American sample.
What we see on client projects is that the loss is not spread evenly. Generic informational queries empty out, because the answer sits above the results. Specific ones, with commercial intent behind them, keep bringing clicks. That is why SEO for zero-click searches shifts towards the questions that lead to an enquiry, not the ones that lead to a definition.
Myth 11: if you're in the top 10, AI will cite you
The same SEOZoom study, on Italian data, takes apart the most common shortcut on the subject. Around 40% of the links cited in Italian AI Overviews come from outside the top ten organic results. Being on page one helps, but it is not the front door people imagine.
Selection, on the other hand, is much narrower than it looks: 175,000 domains cited out of roughly 168 million monitored on Google Italy, or 0.1%. At the top are YouTube with 875,914 mentions and Wikipedia with 864,517, followed by two Italian health sites. That also tells you which sectors this game matters to most: health (40.3% of coverage), finance (20.2%) and family (19.9%). The figures are Italian, but the mechanism they describe, a few sources cited over and over, is not specific to one country.
Getting in is not about a new tag or a tool: it is about having a page that answers a precise question in a form that can be extracted. We have written about it in the guide to generative engine optimisation and in the comparison of AI visibility monitoring tools.
How to check a myth in three steps
The method works for any rule you are offered, including the ones written above.
- Look the rule up in the documentation. developers.google.com/search has a search bar. If a rule concerns ranking and does not appear anywhere, it is not a Google rule: it is someone's habit.
- Look at the SERP you care about. Not the American one if you sell in the UK, not the one for someone else's sector. If the rule says you won't rank under a thousand words and there is a 600-word page in third place, the rule is false, at least there.
- Measure on your own site. Search Console compares two periods. Apply the change to a group of pages, leave the rest alone, look again after six weeks. It is the only way to know whether something works on your domain rather than in general.
The third step is the one almost nobody takes, and it is also what separates a strategy from a list of good practices. If you want to understand which signals really count today, the guide to ranking factors starts from the documentation, not from habits.
What happens when you stop chasing them
Macropix makes custom LED screens in Milan. In 2020, when the project started, the site did not appear for any category keyword in its sector. None of the work that changed that had anything to do with the myths on this page: no keyword density, no scores to raise, no bought links.
What was done instead was the boring work: one page for each product line instead of a PDF catalogue, copy written by people who know the product, and the patience to wait. On "monitor pubblicitario" (advertising display) the site went from position 88, i.e. page nine, to position 2. On "ledwall", 4,400 searches a month in Italy, it is first. In July 2026 the domain held 25.55% share of voice on the ledwall cluster, ahead of amazon.it at 18.08%.
An Italian manufacturer ahead of Amazon on the keywords for its own category is the most practical rebuttal of myth number 1: Amazon's domain authority is in another league, but a page explaining the difference between an indoor and an outdoor LED wall, written by the people who make them, is not. The full story of the project is in the case study on SEO for manufacturing, and the other projects are on the case studies page.
An honest word on timing is needed, since the myth of results in thirty days is the most expensive of all. In its page Do you need an SEO?, Google says only one clear thing on the subject: "No one can guarantee a #1 ranking on Google. Beware of SEOs that claim to guarantee rankings". How long it really takes depends on the site and the competition, and I have set it out on the page about how long SEO takes. If you want an assessment of your own case, that is what we do with SEO projects and, for generative answers, with our AI SEO services.
Frequently asked questions
A final note on where myths are born today. Not on forums and not on courses: in tools. A green score out of a hundred next to a page is a diagnosis without a test result, because no software has access to Google's ranking systems, so every "SEO score" is an opinion presented as a measurement. Whoever set that threshold made an editorial choice, not an observation. The next belief doing the rounds six months from now will almost certainly take the form of a number produced by software, and the way to spot it will be the same as ever: ask where it is written.
