How to Rank in Google AI Overviews Using NotebookLM (SEO Parasite Method)
The NotebookLM parasite SEO method that got Charles Floate cited in an AI Overview in under 24 hours
The new method for showing up inside a Google AI Overview in under 24 hours uses no backlinks or E-E-A-T, just text Gemini spit out. And the source Google cites is a page you own, even if you have no site with any authority on it.
That's the case Charles Floate posted on X. The page his AI Overview pulled from wasn't on his site. It was a public NotebookLM notebook. He didn't break anything at Google, he published a response already shaped like an AI Overview, hosted directly on a Google domain. And Search did what it does constantly: cite itself.
Does this actually work, or is it a one-off nobody can reproduce? The mechanism itself is easy to describe. What's harder to know is how long the window stays open before Google shuts it.
What The Tweet Actually Showed
The query was "best ai seo communities." The Overview returned a structured list and a comparison table. Position 1: AI SEO Rainmakers, Charles's own Skool community. The cited source next to it: "Top 4 AI SEO Communities in 2026, Google Notebook."
Another SEO confirmed in the replies that Google had picked up the page roughly 16 hours after it went live. Charles added 2 things afterward. The step-by-step guide sits behind his paid community. And Google deindexed the page as soon as the tweet started getting traction.
That's the whole case, stripped down. A public NotebookLM page can enter an AI Overview fast. Google can also pull it the moment the trick gets visible enough to notice. You died: the page is removed, and respawn isn't guaranteed.
Why Google Cites Its Own Notebook
Classic parasite SEO means publishing on a domain that already has authority (Medium, LinkedIn, Reddit, a news site, a forum) and borrowing that authority. Here the host isn't some random third-party site. It's notebooklm.google.com.
This differs from a normal borrowed-authority play in 3 ways. The domain belongs to Google, so it gets crawled fast, indexed fresh, and comes with a built-in bias toward citing Google's own properties. The output format matches what an Overview extracts: lists, tables, FAQs, briefings, the exact blocks Gemini pulls from most easily. And the model writing the notebook and the model powering part of Search's generative layer sit in the same family.
You're not optimizing a web page anymore. You're handing Google a source it already trusts, pre-formatted into the shape of its own answer.
What Google Actually Indexes
Not the whole notebook gets crawled. Based on what's been observed on this case, Google mostly picks up the HTML text on the public page: the notebook title, the briefing or report, the FAQ, the study guide, and any table or text output from Studio.
Audio Overviews and Video Overviews don't behave like indexable text. If the goal is a Search citation, the target is content a crawler can read, not a podcast. Feed the crawler a table, not a monologue.
The tables NotebookLM generates matter more than they look. Charles's Overview pulled a name, focus, and price comparison almost as-is. That's the same structure AI Overviews already like extracting everywhere else on the web.
Building A Notebook Worth Citing
Use a personal Google account. Public sharing ("Anyone with the link") is built for personal accounts. Workspace and Education accounts are often locked to internal sharing only. The tool lives at notebooklm.google.com.
Keep each notebook mapped to a single search intent. Don't stack 10 topics in a single notebook. The title should read like a query plus a dated angle: "Top 4 AI SEO Communities in 2026," "Best AI Overview Citation Tools 2026," "GEO Agency Comparison France." In Charles's case, the notebook title became the title of the source cited inside the Overview.
Build the sources like a brief, not a dump. The notebook only invents well from what you feed it. Add a pasted text, written by Gemini, that puts your own offer first. Add the URLs of competitors you're fine citing in positions 2, 3, 4. Add a Google Doc with the ranking, pricing, and positioning if that helps.
The common mistake is uploading 30 generic sources and hoping your brand lands on top. If your name isn't clearly first in the sources, Gemini has no reason to put it first in the output. Garbage sources produce "best of" garbage.
Then force an extractable shape out of Studio: a short intro, a "Top X" list with your brand at #1, a comparison table (name, angle, price, who it's for), a short FAQ, a closing recommendation. An AI Overview doesn't need 2,500 words of blog prose, it needs blocks it can lift as-is.
Getting Google To Find It
Sharing works the same way as a Google Doc. Click Share, set access to "Anyone with the link," copy the URL. A globe icon confirms the notebook is public. Skip this step and the page stays private, invisible to anything Search could treat as a web source.
Fast indexing isn't guaranteed every time. In practice, share the link somewhere (X, LinkedIn, a newsletter), test site:notebooklm.google.com plus the title, wait 12 to 24 hours, then rerun the query in a private window with the target location set. The goal isn't traffic to the notebook itself. The goal is for the Overview to cite that page, and your brand inside it.
If you're testing this on your own visibility across AI search surfaces rather than just Google, getting your brand surfaced inside ChatGPT answers runs on a similar logic with a different platform.
Borrowed Authority, Not Owned Authority
Most "best of" pages take weeks to get indexed and months to look credible enough for an Overview. A public NotebookLM page skips most of that path because it's renting infrastructure Google already trusts instead of building its own from scratch: domain authority that's already Google's, crawl frequency that's high by default, structure that arrives as segmented Studio blocks instead of loosely formatted HTML, and a model writing the source that's the same family as the model reading it for Search.
That's why Charles could say what he said: no links, no E-E-A-T, Gemini-written content, under 24 hours. The site being cited didn't need its own authority. It borrowed Google's, the same way a house guest borrows your Wi-Fi and your fridge.
What You Actually Get
Be precise about the metric here. This tactic doesn't build an asset. It can get your brand to appear inside an AI Overview, speed up a mention on a "best/top/comparison" query, and work as a live test of how an Overview extracts lists and tables.
It doesn't replace a page you actually own, a content cluster under your control, or citations that sit on sources Google can't switch off with a single click. Charles lived that part directly. The moment the trick made noise, the page dropped out of the index. A commenter summed it up well: hacking a Google page you don't control means losing the work the moment they pull it. You're renting a house on Google's land, and Google holds the only key.
Where This Falls Apart
You don't own the URL. That's the whole problem, right there. Google can deindex it, change how public notebooks render, or stop citing them altogether whenever it wants. Public sharing already asks for friction most funnels can't survive too: usually a personal account, often a visitor sign-in with Google on top, nothing close to a normal traffic channel.
Scale it and the mechanic starts looking like exactly what it is. A single test is an observation. 20 notebooks all titled "Top 5 [your offer] 2026" is platform abuse with extra steps, and Google's spam team isn't exactly known for its sense of humor about that.
A citation isn't a click either. The Overview can lift your ranking without sending a single session to your site. You walk away with a mention, not a conversion. And if the notebook is nothing but your offer dressed up as a neutral ranking, the Overview might skip it outright, or Google might flag the pattern once it's exposed at scale.
You're not building a moat. You're borrowing one for the weekend.
Using It Without Fooling Yourself
The sane version of this isn't "replace SEO with NotebookLM." Pick a single informational query with a strong list or table format. Publish a clean public notebook with real competitors sitting around your own offer. Check whether the Overview cites it within 48 hours. Then copy the winning structure onto your own site. Treat the notebook as a disposable prototype, not a ranking you plan to defend.
If GEO and AI search visibility are new territory for you beyond this single trick, the bigger mistakes killing your AI search visibility covers the wider pattern this fits into.
NotebookLM is mostly a good sandbox here. It uses the same RAG architecture behind a lot of generative answers, with explicit citations, tables and FAQs already in the shape Search likes to extract. The parasite version is just the aggressive read of that observation.
Charles didn't break Google. He rented a room in Google's own house and got Search to point at it. It works fast. You don't own it. And Google already showed it knows exactly where the light switch is.
Sources
- Charles Floate's original post on X documenting the NotebookLM AI Overview citation
- NotebookLM sharing and publishing documentation at notebooklm.google.com
This post may contain affiliate links. If you click them, I might earn a small commission — costs you nothing, and helps me keep shipping quality articles every day for your reading pleasure.
If you're shipping AI products, know the difference between a viral hack and something that survives in production. The demo-vs-product checklist in the welcome kit covers the 8 criteria Google uses to decide what stays indexed and what gets deindexed once it gets attention.