Pular para o conteúdo
← Back to Skalablog

Published article

5 Parts of an AI SEO System That Still Gets Clicks

Products & BusinessChatGPTGeminiOpenAI

An AI SEO system earns clicks by getting your brand named inside AI answers, not just by ranking pages. The version described here combines AI-written posts, Google Search Console data, and off-site brand mentions, and one site run this way reportedly pulled 27,800 clicks in 90 days.

What an AI SEO system is and where the 27,800 clicks came from

An AI SEO system is a workflow where AI tools write and publish content, your own Search Console data chooses the next keyword targets, and off-site brand mentions teach AI engines to name you in their answers. In the system described by SEO practitioner Julian Goldie in October 2026, one small site in a network of AI-published sites reportedly drew about 27,800 clicks in 90 days with no hand-written posts.

Two caveats belong next to that number. It is a speaker-reported result from his own site network, not an independently audited case study, and the traffic reportedly climbed only after a long flat period where the site sat near 81 clicks a week before reaching 405. Treat the figure as one operator's first-hand experience rather than a guaranteed outcome.

Why ranking number one stopped being the whole game

Ranking first no longer guarantees traffic because search engines increasingly answer the question on the results page itself. Google's AI Overviews, which generate an AI summary above the normal results, are the most visible version of this shift.

The video cites three data points that describe the shift, each from a different measurement source:

  • BrightEdge, an enterprise SEO platform, tracks AI Overviews appearing on roughly 48% of the searches it monitors, a share it reported as up 58% year over year, and near 80% in education-related queries.
  • SparkToro, audience research firm founded by Rand Fishkin, together with data provider Similarweb, measured that about 60% of US Google searches ended without a click in 2024, and the video attributes a rise to roughly 68% in early 2026 to their ongoing tracking. That second figure is the speaker's reading of their data, so verify it against their latest zero-click research before quoting it.
  • Ahrefs, the SEO toolset company, published research in 2025 finding that the top result's click-through rate drops when an AI Overview appears; the video cites that drop at about 58% across hundreds of thousands of keywords, so check the exact number in the Ahrefs study it references.

The direction of travel is consistent across sources even where exact percentages differ: when an AI summary answers the query, fewer people click anything.

The everywhere engine: one keyword, several sites, same day

The publishing flywheel is the first component. Instead of one site posting one article, the operator feeds one keyword and one real example from his own work into an AI writing tool, which produces several different posts with different titles and angles. These publish across multiple small sites the same day, interlink where relevant, and get pinged so Google crawls them quickly.

Each published post generates fresh Search Console data, that data reveals new keyword opportunities, and those feed the next round. This loop is what the operator calls a flywheel: slow to start, then self-reinforcing.

Be clear about what this component is. It is one marketer's operating method, shared as 'this is what works for me, not a recommendation'. Running multiple sites to cover one topic carries risk if the content is thin or duplicative, and Google's guidance penalizes scaled content abuse regardless of whether a human or a model typed it.

Using Search Console gaps and leaks to pick targets

Google Search Console, Google's free dashboard showing how your site performs in search, is the fuel source. It is the only keyword data source that shows queries only your site saw, and in June 2026 Google added a dedicated AI performance report inside it that shows how often your site appears in AI-driven results.

The method sorts opportunities into two categories:

  1. Gaps: queries where your site appears far down the page with impressions but no purpose-built page. The video's example: a keyword with 650 impressions at position 21 and zero clicks, where no real page existed. The fix is to build the page.
  2. Leaks: queries where the page ranks well but the snippet fails to earn the click. The video's example: over 14,000 impressions at position 8 with under 1% click-through. The fix is rewriting the title and content, not building anything new.

Fixing leaks first is the fastest win because the ranking already exists. This part of the AI SEO system requires no tool purchases and works on any site with Search Console access.

Quality control: information gain and the 13-rule checklist

Google does not ban AI-generated content; its guidance targets low-value scaled content. The safeguard in this system is a written rule set the AI follows on every post. The operator's checklist runs about 13 steps, with the core requirements being: every fact traceable to something that actually happened, a real example near the top of each post, titles written to earn the click, and a clear next step at the end.

The distinguishing ingredient is what he calls information gain: adding first-hand results, tests, and before-and-after data that exist nowhere else on the web. Because that material came from your own work, competitors cannot copy it. If you can describe real outcomes from your job, you can supply it; that is the stated skill floor for the whole system.

Brand mentions outrank your own site's authority

What others say about you moves AI answers more than what you publish about yourself. The video cites MP Digital, a digital marketing agency, which studied 82 factors behind whether ChatGPT, OpenAI's assistant, recommends a brand: relevance ranked first, and brand mentions across the wider web ranked second, above the brand's own site authority.

That finding explains why this AI SEO system extends beyond your own properties. The operator describes building and posting in a subreddit that drew over 244,000 views in 30 days, so that when AI crawlers scan the web they repeatedly encounter the same brand next to the same topics. The claim to carry forward is directional: mentions on sites you cannot edit appear to influence whether assistants like Perplexity, the citation-backed AI search engine, and Gemini, Google's assistant, name you in answers.

Fewer clicks, better clicks: the conversion evidence

The traffic you lose to AI answers is disproportionately the low-intent traffic. Adobe analyzed over a trillion visits and reported that visitors arriving from AI referrals were about 42% more likely to take an action such as a purchase or signup, and spent 48% more time on the pages they landed on, per Adobe's 2025 AI traffic analysis. A smaller study of a few hundred companies, cited in the video, found AI-referred visitors converted about five times more often than visitors from traditional search.

The practical reading: AI answers act as a pre-filter that screens out casual browsers. If your measurement stops at the sessions count, you will read this shift as a loss. Measure conversions per visit alongside total traffic before deciding whether zero-click search hurt your business.

FAQ

  • Does Google penalize AI-written content? Google's stated position targets low-value scaled content, not the writing method. The video's own evidence is about 1,500 AI-assisted posts across its site network with traffic growing throughout, which is speaker-reported experience, not a Google guarantee.
  • Do I need technical skills to run an AI SEO system? The described setup publishes through an API without dashboards or manual logins, and the operator's stated input is one keyword plus one real example per cycle. You still need to supply genuine first-hand experience, which no tool can generate for you.
  • How long until results appear? Expect a flat period first. One site in the reported network sat near 81 clicks a week before climbing to 405, and the 27,800-click result came only after that flat stretch. New domains typically wait months before Google trusts them.
  • What should I do first if I try this? Write down three to five real results from your own work, open Search Console's AI performance report added in June 2026, and fix click-through leaks on pages that already rank before building any new content.
  • Is the 27,800-click figure independently verified? No. It is the speaker's first-hand account of his own site network. The external statistics on AI Overviews, zero-click search, and AI-referral conversion come from BrightEdge, SparkToro, Ahrefs, and Adobe respectively.

Source video