On 1 October 2026, Google updated its guidance on AI content and helpful, people-first content. Buried in it is a line that matters more than it looks, for search and for AI search alike. Google now spells out four things its quality raters assess in the main content of a page: Effort, Originality, Talent or Skill, and Accuracy.
The first one is the interesting one. Google defines effort as "the human effort put into creating the content, or the system that produces it." In other words, how much real work went into this, and can anyone tell.
This is not Google turning against AI
It is worth being clear, because the headlines will get this wrong. Google is not penalising content for being made with AI. Its position is that the tool does not decide quality, the process does. Its own words: "it is critical to manually factcheck and review all AI-generated content" before it goes live.
What Google is penalising is low effort. Mass-producing pages "without manual oversight and curation" is, in Google's framing, exactly that, and it violates the guidance. The September 2026 spam update, which finished rolling out on 24 September, went after precisely this: near-identical scaled content, pages built only to rank, fake author profiles with AI-generated photos and invented credentials. All of it got demoted.
What got rewarded was the opposite: human oversight, proprietary data, real expert authorship, original analysis, fact-checked information, genuine company knowledge on the page.
Why this matters for AI referral, not just Google
Here is the part that is ours, not Google's. The same quality bar that decides a Google ranking increasingly decides whether an AI assistant will cite you.
When ChatGPT, Claude, Gemini or Perplexity answer a buyer's question, they pull from pages they can retrieve and trust. Thin, templated, clearly mass-generated content is the easiest thing in the world for them to pass over, and it is getting easier, because the systems that detect the "template feel" and conceptual thinness of low-effort content are the same systems feeding the answers. Substance, originality and evident expertise are what get a page lifted into an answer and attributed to a source.
So perceived effort is now a signal on two fronts at once. A page that reads as real work, written by someone who knows the subject, with specifics only that business would have, gets both ranked and recommended. A page that reads as filler gets neither.
The shortcut that is now a liability
For two years the pitch doing the rounds has been volume: generate hundreds of AI pages, flood every keyword, let the algorithm sort it out. After 1 October, that is not a neutral bet that might not work. It is an active liability. Scaled, low-oversight content is what the spam update demotes, and it is what AI assistants skip. Businesses that took that route are now spending effort unwinding it.
The uncomfortable truth is that the cheapest way to produce content is now the one most likely to cost you visibility.
What actually earns the citation
None of this is complicated, it just takes effort, which is the whole point.
Fewer, better pages. One genuinely useful answer to a question your customers really ask beats fifty thin variations of it. Depth, specifics, and a point of view the competition does not have.
Your own knowledge, on the page. The things only you know, your real numbers, your actual process, the cases you have seen, are exactly what Google calls originality and expertise, and exactly what an AI assistant has nowhere else to find. That is what makes you citable.
Real authorship and accuracy. A named, credible author. Claims that are checked. No invented credentials, no borrowed stock expertise. These are now scored directly.
Human oversight of anything AI touches. Using AI to help is fine, Google says so. Shipping its first draft unreviewed is not. The value you add before you publish is the effort being measured.
This is the same conclusion from a different direction: the businesses AI recommends are the ones that put real work into being worth recommending. Google has now made that work a named, measurable signal. The effort you can see is the effort that gets rewarded.