Common SEO myths

SEO has an unusually long memory for ideas that were never true, or stopped being true years ago. Some persist because they sound plausible, some because a third-party tool named a metric that people assumed Google must use too, and some because a tactic once worked and the folklore outlived the algorithm. This page collects the most common ones and gives each a plain correction, drawn from Google’s own statements, its documentation, or the ranking systems exposed in the 2024 Content Warehouse leak.

A note on scope: these corrections describe Google’s systems. Other search engines and AI answer engines have their own retrieval logic.

Is there a Google Sandbox?

Not as a formal mechanism. Google has denied that it runs a deliberate “sandbox” that holds new sites back for a fixed period. What people observe and label a sandbox is real, but it is the ordinary consequence of a new site having little established trust: Google has less data to assess it, so rankings are volatile and often start low before settling. The effect looks like a penalty box; the cause is simply the absence of a track record. The fix is not to wait out a timer but to build the signals, links, mentions, and useful content, that let Google evaluate the site with confidence.

Is Domain Authority a Google ranking factor?

No. Domain Authority (Moz) and Domain Rating (Ahrefs) are third-party metrics invented by SEO tools to estimate a site’s strength. Google does not calculate them and does not use them. The honest nuance is that “Google does not use vendor scores” and “Google forms no view of a site as a whole” are two different claims, and only the first is settled: the 2024 Content Warehouse leak describes a siteAuthority field within its quality signals, and the endpoint exploit disclosed in December 2024 described a per-subdomain site quality score.¹ A leaked field name is strong evidence that something exists, not confirmation of how it is weighted or whether it is live. Treat DA and DR as convenient competitive proxies, not as something Google reads. Domain authority explained works through the distinction in full.

Is there a duplicate content penalty?

No. There is no penalty for duplicate content in the ordinary sense. When Google finds duplicate or near-duplicate pages, it filters and canonicalises: it picks one version to show and consolidates signals onto it.² That can mean a page you wanted to rank does not, but it is a selection process, not a punitive action against your site. The exception is content duplicated deliberately and at scale to manipulate rankings, which falls under spam policies, a different thing from having a printer-friendly page or some boilerplate repeated across a site.

Is there an optimal keyword density?

No. There is no target percentage of times a keyword should appear, and there never was a threshold that Google rewards. What the myth gets wrong is the ratio, not the relevance of the words themselves: sworn testimony in the US antitrust case describes topicality as built from ABC signals, where B is the terms in the document, so the words on the page plainly still count.³ What does not exist is a density figure that rewards you for hitting it. Retrieval also matches meaning as well as strings, so a page can rank for queries it never states verbatim. Writing to hit a density target tends to produce worse, more repetitive copy, which is the opposite of the goal.

Are LSI keywords real?

No. “LSI keywords” is a persistent misnomer. Latent Semantic Indexing is a 1980s information-retrieval technique unrelated to how Google matches meaning today, and Google’s representatives have flatly said there is no such thing as an LSI keyword. What is real, and worth doing, is covering a topic comprehensively with the related terms and concepts a genuine expert would use. That is semantic completeness, not a magic list of “LSI terms” to sprinkle in.

Is bounce rate a ranking factor?

No. Google does not use Google Analytics bounce rate as a ranking signal; it does not have access to your GA data for ranking purposes, and has said as much repeatedly. This is worth separating from the click-and-satisfaction behaviour Google does model in its own systems. Whether a searcher clicks a result and is satisfied is part of ranking; your Analytics bounce-rate figure is not the same measurement and is not what feeds it. The click signals Google does model are aggregated across many searchers and are not readable from your own analytics at all.

Is the meta description a ranking factor?

No. The meta description does not influence ranking. Its job is to inform the snippet Google may show in results, which affects whether people click, not where you rank. That still makes it worth writing well, because click-through matters commercially, but it belongs in the “improve the snippet” column, not the “improve the ranking” column. Google also frequently rewrites descriptions to match the query, so treat yours as a strong suggestion rather than a guarantee.

Do hyphens in a domain name hurt rankings?

No. Google has said its algorithms do not treat a hyphenated domain as low quality, and there is no filter that demotes domains for containing a dash.⁴ The belief dates to a “hyphen filter” that SEOs described in 2004 and that Google never ran. The real objection is to a different thing wearing the same clothes: keyword-keyword domains like cheap-blue-widgets.com. John Mueller has said he is not a fan of them, not because of the punctuation but because they carry spam associations, they age badly, and they trap a business that later changes what it sells.⁴ A hyphen in a genuine brand name costs you nothing in ranking. It is simply harder to say out loud.

Does title tag length affect rankings?

No, and this one deserves care, because the correction is often overstated into a second myth.

Google sets no limit on how long a <title> element can be, and truncates the displayed title link to fit the device width.⁵ There is no ranking penalty for a long title, and Whitespark’s 2026 survey of local search practitioners scored title-tag length near the bottom of its factor list. As a ranking lever, the 50 to 60 character rule is folklore.

But the rule survives for a reason that is not folklore: length is the strongest predictor of whether Google rewrites your title. Zyppy’s study of 80,959 titles found those of 51 to 60 characters were rewritten least often, while titles over 70 characters were rewritten 99.9% of the time.⁶ The character budget does not buy you rank. It buys you authorship of your own search result, which is worth having.

A related claim travels with this one and does not survive contact with a source: that each additional word dilutes the ranking weight of the others. Google has never said it, and the study most often cited for it does not claim it. Front-load the important words because truncation eats the tail, not because of a weighting effect nobody has demonstrated.

Does domain age help rankings?

No. Google’s John Mueller has said plainly that domain age helps nothing. An older domain often correlates with better rankings, but the cause is what accumulates over time, links, content, and trust, not the age of the registration itself. Buying an old domain does not buy you a ranking advantage from the age; any benefit comes from the history and signals attached to it, and an aged domain with a poor or unrelated history can be a liability rather than a head start.

Are Core Web Vitals a primary ranking factor?

No, they are a minor one. Core Web Vitals and the broader page experience signals are real and worth improving, but Google has been consistent that relevance and quality win: a highly relevant result with mediocre page experience will still outrank a fast, pretty page that answers the query less well. Page experience is best understood as a last-mile tiebreaker, the thing that can separate closely matched results, not the lever that lifts weak content. Fix your Core Web Vitals, but not before you have fixed relevance and quality.

Does a canonical tag guarantee the chosen URL?

No. A canonical tag is a hint, not a command. It tells Google which version you prefer, but Google can and does choose a different canonical, which shows up in Search Console as “Duplicate, Google chose different canonical.” If Google keeps overriding your canonical, the fix is usually to remove the conflicting signals, inconsistent internal links, sitemap entries, and redirects that point elsewhere, rather than relying on the tag to force the decision.

Do more indexed pages mean better rankings?

No. A larger index is not inherently better, and can be worse. Publishing many thin or near-duplicate pages to inflate the count is index bloat: it dilutes crawl attention and can drag on how Google assesses overall site quality. What matters is that your valuable pages are indexed and your low-value ones are not. Fewer strong pages beat many weak ones.

Are social signals a direct ranking factor?

No, not directly. Google has said it does not use social media shares, likes, or follower counts as direct ranking signals. Social activity can help indirectly by putting content in front of people who then link to it, cite it, or search for the brand, all of which feed signals Google does use. So social matters for distribution and demand generation, just not as a dial Google reads off the platforms.

Does only the first link on a page count?

Not as a fixed rule. The old “first link priority” rule held that when a page links to the same URL more than once, Google counts only the anchor text of the first link and ignores the rest. It had a kernel of truth, Matt Cutts said in 2009 that Google often takes the first anchor, and a kernel is all it needs to survive as folklore. But the absolute version, that the first link always wins and every later one is wasted, is not something Google has confirmed. Asked directly in 2021 whether it used the first or the longest anchor, John Mueller said Google has not defined that at all, does not naively take the first link, and instead tries to understand the site structure the way a user would, adding that the behaviour can vary over time.⁷ Controlled testing supports the softer reading: Google is selective rather than strictly first-link, counting some anchors to a given URL and not others.⁸ The kernel worth keeping is that duplicate anchors to the same page do not each pass their own signal, so linking one page five times with five keyword anchors will not earn five anchors’ worth of weight. The practical upshot is unchanged: link naturally, and do not engineer link order to game an anchor rule Google has said it does not run.

Do links from “blacklisted” domains cause a penalty?

Mostly no. There is no official Google blacklist of domains that infect you on contact. Google’s stance on low-quality inbound links is largely to ignore them rather than penalise you for links you did not create, which is why negative SEO via toxic links is far less effective than it once was. Bad links tend to carry no benefit rather than active harm. The disavow file remains available, but for most sites it is unnecessary; reserve it for cases tied to a manual action.

Do you need to do anything special for AI search?

The myths above are the classical set, accumulated over two decades. A newer set has formed around AI search in a fraction of the time, and Google has published its own list of them under the heading “Mythbusting generative AI search: what you don’t need to do”.⁹ Its five, in Google’s framing and applying to Google Search only:

• llms.txt and other “special” markup. “You don’t need to create new machine readable files, AI text files, markup, or Markdown to appear in Google Search (including its generative AI capabilities), as Google Search itself doesn’t use them.” Google adds that maintaining one for other services “will neither harm nor help your site’s visibility or rankings in Google Search”. See llms.txt for where it may still have value elsewhere.
• “Chunking” content. “There’s no requirement to break your content into tiny pieces for AI to better understand it”, and “there’s no ideal page length”.
• Rewriting content just for AI systems. “AI systems can understand synonyms and general meanings”, so you need not chase every long-tail variation.
• Seeking inauthentic mentions. Google’s ranking systems focus on high-quality content while separate systems block spam, and “our generative AI features depend on both”.
• Overfocusing on structured data. “Structured data isn’t required for generative AI search, and there’s no special schema.org markup you need to add”, though it remains worth using for rich result eligibility.

Two cautions on how to read that list. It is Google speaking about Google, so it settles nothing about how ChatGPT, Claude or Perplexity retrieve, and those systems have their own documented behaviours. And “not required” is not “worthless”: structured data and clear structure remain good practice for reasons that predate AI search.

One more myth belongs here because it is repeated confidently and is untrue: AI-generated content does not carry an automatic penalty. Google’s position is about quality rather than production method, and using automation “to generate content with the primary purpose of manipulating ranking” violates its spam policies, which is a narrower claim than the one usually reported.¹⁰ The risk lies in publishing unhelpful content at scale, not in the tool used to write it.

A final one to watch, because it illustrates the whole pattern: a claim circulated widely in 2026 that Google had tightened the Core Web Vitals thresholds. The documented thresholds are unchanged. Any specific new figure should be checked against Google’s documentation before you act on it, which is exactly the test the next section describes.

The pattern behind the myths

Read together, these corrections rhyme. Most myths prescribe a shortcut, a density figure, an aged domain, a canonical that forces Google’s hand, a page count, and the correction is almost always the same: the shortcut does not exist, and the thing that actually works is unglamorous. Cover topics properly, keep the site technically sound, earn genuine references, and give searchers what the query wanted.

When a new “rule” surfaces, the test is two-part. First, does it describe something Google has actually said it does, or something a tool or a blog inferred that Google must be doing? Second, and this is where the Sandbox and Domain Authority entries above are instructive, does the disclosed record complicate the public statement? In both cases Google’s public denial was narrower than the internal picture the 2024 leak revealed. A public statement is strong evidence, not the whole of it.

Footnotes

1. Google’s site quality scoring system revealed through endpoint discovery — PPC Land ↩

2. Demystifying the duplicate content penalty — Google Search Central ↩

3. The ABCs of Google ranking signals: what top search engineers revealed — Search Engine Land ↩

4. Google Says Hyphens In Domain Names Not Considered Low Quality — Search Engine Roundtable ↩ ↩²

5. Influencing your title links in search results — Google Search Central ↩

6. Google Rewrites 61% of Page Title Tags — Zyppy ↩

7. Google: Multiple Anchor Text Links To Same URL On Same Page, The First Or Longer Doesn’t Matter — Search Engine Roundtable ↩

8. How Google’s Selective Link Priority Impacts SEO — Zyppy ↩

9. AI features and your website — Google Search Central ↩

10. Google Search and AI-generated content — Google Search Central ↩