Learning from the leaked Google ranking documents with Gemini 1.5 Pro
Вставка
- Опубліковано 19 лис 2024
- Documents describing Google's ranking systems in great detail have been leaked by someone who sent an email to Rand Fishkin. In this video, I share my thoughts on Rand's post and then at around 35 mins or so I play around with analyzing this document with Gemini 1.5 Pro in AI Studio.
Hope you enjoy this type of video. I share a lot of my thoughts (perhaps a bit to rambly! I had to think a lot here) on how machine learning systems work to use all of the many attributes described in this documentation. This includes links, clicks, return to search results and much more.
Take home point: Google uses MANY signals/attributes/features in their calculations that predict what is likely to be helpful to the searcher. Our goal should be to create content that people choose to engage with and ultimately find to be the satisfying answer to their query - or in other words, HELPFUL.
Funny story - the background behind me is AI generated in Google meet. You can generate them with a prompt now. I asked for a Panda and Penguin on a shelf and instead it hung a stuffed Panda on the wall!
Links mentioned
Marie’s newsletter: mariehaynes.co...
Marie’s notes (new - paid newsletter - follow along as I take notes on things like this each week.)
community.mari...
Marie’s Course and book: SEO in the Gemini Era is out soon.
The Docs referred to in this video: hexdocs.pm/goo...
Rand Fishkin’s post. He received an email with the contents of the leak. sparktoro.com/...
Mike King’s article on these documents:
ipullrank.com/...
Thank you for a detailed video about this leak. I'm looking forward to read your future blog posts about it.
Is there a link to the actual documents themselves?
I wish you’d update your new creating helpful content workbook we bought. It references Google documents that no longer exist. The very first question in the workbook references a document that is now redirected and does not include the text your workbook asks us to provide. Frustrating.
This explains why EAT has never worked.
Errr ahhh errr ahhh h errrr
You're saying "uuuhhh" roughly every 6 seconds