Google added a definition of main content, four attributes for judging it and a warning about fabricated creator profiles to its guidance page on helpful, reliable, people-first content, according to a video published on October 2, 2026 by SEO consultant Marie Haynes.

In Short

Google edited a help page that explains what good web content looks like, adding a plain description of what counts as the main part of a page and four traits - effort, originality, skill and accuracy - used to judge it. It matters to anyone whose website depends on visitors from Google, because the same edit says invented writer names, computer-made headshots and false credentials make a page look untrustworthy. Nothing in the material examined says rankings changed with the edit; what changed is the written explanation of what Google's systems and human reviewers are meant to look for.

What Haynes reported, and what the capture shows

Haynes, who runs the "Search News You Can Use" video series, published the 27-minute-7-second walkthrough on October 2, 2026. According to the YouTube page attached to the material, it carried 4,213 views, 90 likes and 7 comments, and is divided into 16 chapters. In the recording she described the changes as the part of the documentation that "changed today", which places the edit at or just before the video's date. Neither the video nor its description gives a more exact time.

The page in question is "Creating helpful, reliable, people-first content". The capture of the Google page that accompanies the video shows only its opening portion, which says the company's automated ranking systems are "designed to prioritize helpful, reliable information that's created to benefit people". It then points creators to a self-assessment, suggests asking trusted outsiders for an honest assessment and recommends auditing any drops in traffic. The new sections are not visible in that capture. Their wording is therefore known only through Haynes reading them aloud, a limit worth keeping in mind throughout.

The video description lists four Google documents as the basis for the discussion: the helpful content page, the Search Quality Rater Guidelines, Google's spam policies and the "How Search Works" ranking explainer.

How Google now defines main content

According to Haynes, the new passage says Google defines main content as "any part of the web page that directly helps the page achieve its purpose". She linked the concept to the Search Quality Rater Guidelines, saying the first task raters receive is to work out a page's purpose. The January 23, 2025 edition of those guidelines, which PPC Land hosts as a PDF, lists the topic as section 2.4.1, "Identifying the Main Content (MC)".

Haynes listed the elements the new text counts as main content:

  • primary text and media, including articles, images, audio and video;
  • interactive features such as calculators and tools;
  • user-generated contributions such as comments;
  • tabbed or expanded sections;
  • page titles and headings.

She drew practical readings from several of these. A mortgage calculator, in her example, belongs at the front of a page rather than behind four paragraphs of explanation. Comments need not sit at the top, but where they are what helps visitors, they should be easy to find. On tabs, she said Google's text makes clear that content behind them is acceptable, and that a page is not penalised for it. She cited a client whose product video reviews sat behind a tab labelled "additional information" that nobody clicked; the reviews were moved out. She also said she uses Microsoft Clarity to check whether visitors open tabs.

Page titles drew an aside. Because vector-based search retrieves pages that sit mathematically close to a query, she said, a title carrying the main topic plus a couple of related topics may help machines, though she stressed that the primary audience is the human reader. The existing self-assessment questions already ask whether a page avoids excessive ads and lets visitors tell main content from the rest of the page.

Four attributes: effort, originality, talent or skill, accuracy

The second addition lists what Haynes called the fundamental attributes raters are trained to evaluate in main content. Each, per her reading, carries a short definition.

Effort is "the extent to which human work went into creating the content" or the systems behind it. Google's examples of high effort, as Haynes read them, are writing original analysis, translating a poem manually and building a custom interactive map for a game. Little to no effort covers automatically generating pages from feeds, or using generative AI to produce large amounts of text without manual oversight or curation. Haynes noted that the word "effort" appears 120 times in the rater guidelines, the same count that PPC Land's July 2026 coverage cited from her analysis of unindexed pages. She read the text as accepting AI assistance so long as human work accompanies it.

That contrast sits next to Google's scaled content abuse policy, which, when introduced on March 5, 2024, applied to both human-written and AI-generated pages. Rater treatment of reused material has tightened too: PPC Land reported that the guidelines separate a Low rating, where some main content is reused with minimal curation, from a Lowest rating, where almost all of it is copied or paraphrased with no added value.

Originality asks whether content offers unique information or perspectives not already available on other websites. Haynes said she runs a saved prompt in the Gemini sidebar in Chrome whenever a site declines after a core update, to judge whether pages look like what Google calls commodity content. Almost every time, by her account, little original value turns up, and what does exist often fails to meet the searcher's need. She added that content teams producing at volume find it hard to supply material that does not already exist online.

Talent or skill covers clear writing, well-produced video or functional page tools, and Google's text adds that not all content requires specialised expertise. The example read out was a person sharing a way to clear a driveway of snow. Haynes tied this to a forum about living with liver cancer that the rater guidelines rate highly: contributors are not doctors, yet where a page's purpose is to support people going through the illness, lived experience is what counts. A page explaining the disease and its treatment would need medical expertise instead.

Accuracy applies stricter standards to topics that could significantly affect people's lives or well-being, where content must be highly accurate and consistent with established expert consensus. Haynes traced that emphasis to the August 1, 2018 core update, known as Medic, which changed how medical content was evaluated. She added that Google changed its recommendations on using AI in content on the same day, to stress verifying accuracy; that page is not part of the attached material.

A warning on deceptive authorship

Google's page already asked whether pages carry a byline where one might be expected. Haynes was blunt that bylines are not a ranking factor, adding that she does not think a line in Google's algorithm awards points for an author bio, though readers tend to like them. A site such as IBM, where every page plainly comes from the company, may not need one. According to her reading, Google says that clearly indicating who created content likely aligns with the concepts of E-E-A-T and strongly encourages accurate authorship information where readers would expect it. Haynes dated the arrival of the second E, for experience, to December 2022.

The new sentence, which she flagged as the part that is new, runs the other way. Fabricating creator profiles, whether through AI-generated headshots, made-up names or false credentials, so that content appears to be written by human experts is "a form of deception". Per Haynes's reading, Google adds that any such deception makes a page untrustworthy to both users and its automated quality systems and marks the page as weak. She noted that "trust" appears 191 times in the rater guidelines, against 120 for effort, and said Google values pages it can trust.

Does the sentence create a new penalty? Nothing in the material says so. It describes how Google regards such practices; it does not describe a detection mechanism or a consequence.

Haynes's reading: models trained to replicate the page

Much of the video is interpretation rather than reporting. Haynes argued that the document matters because Google's ranking now leans on machine learning. The helpful content system, introduced in mid-2022 shortly before ChatGPT appeared, was in her telling the first strong use of such models in ranking, and she recalled that people at the time could not see how the questions in the document could become an algorithm. In March 2024, she said, Google merged that system into core ranking, calling it an evolution in how helpfulness is identified. PPC Land's tracker coverage records that the March 2024 core update ran 45 days and absorbed the helpful content system, the longest rollout in that dataset.

She also described NavBoost, which she said stores each query along with the actions taken: a click on the first or third result, a return to the results page, or no return at all. Her view is that such behaviour data tests whether a ranking decision was right and helps the overall model predict what searchers find helpful, rather than changing the signals attached to a single page. She went further, suggesting Google may now need user signals less than before, because its models learn to replicate what the document describes. That is her opinion, and Google has not said so in the material examined.

Nor, she stressed, does Google run a checklist scoring "one point for original content". The questions describe a target that models try to approximate.

Timing was the speculative thread. Haynes said Google tends to publish major documentation changes shortly before a major update, and linked this to the new Gemini 4 Argon foundation model, which she said Google had announced but not released to the public. She recalled that Gemini 1.5 arrived just before the March 2024 core update. She said she expects another core update soon, then hedged that it might not happen. She does not think the content is new in Google's algorithms, only that the edit could be a way of preparing the ground.

The spam update running in the background

Haynes recorded the video while the September 2026 spam update was still rolling out. According to PPC Land's coverage, it began at 09:15 Pacific on September 24, carried an allowance of up to 14 days, and was the fourth spam update of 2026. The three before it each finished in about three days; the June update ran from June 24 to June 26. Spam updates refresh Google's automated detection, known as SpamBrain.

Haynes said she highly suspects the September update will reduce the value of content generated by AI in bulk. Using AI to help with some pages is no worry, in her view; using it to produce hundreds or thousands of pages that would not exist without search is a concern. Several sites she has monitored for a long time saw large gains, she reported, though she could not say whether that reflected spam being cleared from their verticals or their own improvements. She also said that every page in Google's systems carries a spam score, a claim she attributed to an earlier article of hers.

Her July 16, 2026 analysis, covered by PPC Land, tied a rise in unindexed pages to commodity content and suspected the June spam update of hitting sites producing it at scale, while saying she could not yet prove the link. Whether the October 2 edit relates to either update is, again, her inference.

Why the edit matters to publishers and marketers

Haynes said everybody is seeing traffic declines because of AI answers. Ahrefs data reported by PPC Land put the cut for top-ranked pages at 58% as of December 2025. Her argument follows: if a searcher already has the facts in an AI answer, a page needs something beyond them to earn a click, and the effort, originality and accuracy attributes spell out what that something might be. Google had already pushed the same line in May, when PPC Land reported that its AI search guide stressed non-commodity content, and when it extended its spam policies to AI Overviews and AI Mode on May 15, 2026.

For paid media teams the relevance is indirect but real. A site that falls foul of spam enforcement can lose more than organic traffic: since December 2024, Google Ads has disapproved ads pointing to destinations removed from Search through a manual action. Publishers living off search referrals face the sharpest exposure, since lower visibility feeds directly into impressions available to sell. Haynes's own earlier work, also covered by PPC Land, found sites hit by the helpful content system recovering after the June 2025 core update, a reminder that the same quality criteria cut both ways.

She closed with an anecdote. Years ago she reviewed a well-written article of roughly 400 words about garage refrigerators and still could not tell what one was. The page Google ranked looked spammy, yet an image at the top showed an ordinary refrigerator with extra controls, and that met her need. Her conclusion was that every page should be built around the need of the person who lands on it.

What the material leaves open

Several points cannot be settled from the attached documents. The exact wording of the new sections is secondhand. The exact time of the edit is unstated. Google has not, in anything supplied, linked the edit to a coming update or to the September spam run, and Haynes herself said she does not think the content is new in the algorithms. Her counts of 120 and 191 mentions are hers; PPC Land's July coverage repeated the first without independent verification. And whether Google's systems can actually tell a fabricated creator profile from a real one is not addressed.

Timeline

Summary

Who: Google, whose Search documentation team maintains the page; Marie Haynes, the SEO consultant who reported and interpreted the changes; publishers, SEO teams and brands whose pages depend on Google search traffic.

What: Per Haynes's video, Google added to its helpful, reliable, people-first content page a definition of main content, four attributes raters use to judge it - effort, originality, talent or skill, and accuracy - and a warning that fabricated creator profiles amount to deception.

When: Haynes published the video on October 2, 2026 and said the changes were made that day. The September 2026 spam update, begun on September 24, was still rolling out.

Where: In Google Search's public documentation on creating helpful content, with reach across organic search, the rater guidelines it draws on and, indirectly, Google Ads eligibility rules for penalised sites.

Why: Haynes reads the edit as a description of the targets Google's machine learning ranking systems are trained to approximate, and as possible groundwork for a coming update. Google has not confirmed either reading in the material examined.