This 👇
One cause I often observe is that B2B websites use language that is too broad to target B2B searches. In addition, they may use brand recognition and visuals to signal that their target audience is B2B, rather than saying so explicitly. https://t.co/TOATWg2WW2
🚨 Google is finally releasing a Search Generative AI performance report in Search Console! It doesn’t seem to imciude prompts / topics information or clicks data but … it’s a start!
“The new Search Console reports are designed to give you dedicated views of your impressions within generative AI features on Search, such as AI Overviews and AI Mode, as well as generative AI features in Discover”
They’re initially rolling them out to a sub set of sites for testing purposes.
Can’t wait to test. Check the announcement 👇 https://t.co/DVs3x7P4So
@IsabellaMWeber 1. Arrogance and short term profit thinking is a big problem in the West, not just in Germany.
2. Chinese definitely stole tons of IP, not just from the Germans.
3. China has racism and nationalism aswell and doesn’t allow immigration like the West does.
Three independent sources said the same thing: the GEO playbook doesn't work.
Frontier labs have already publicly hedged on what their systems can do. Ahrefs ran a controlled study and found no citation uplift. Last Friday, Google published documentation saying the prescriptions aren't needed.
The frameworks keep selling.
New on The Inference: Mt Stupid has a pricing page. 🔗👇🏻
https://t.co/MFcszV3NDO
@pedrodias Great post - again!
I will say tho the GEO hype has done 2 things:
1) help get the word out that Google’s max rendering capabilities are not the be-all and end-all.
2) motivate clients to write more clearly and precisely about their offers.
💃 SEO has a new party dress🕺🏼
Don’t let your clients ChatGPT SEO advice frustrate you. Welcome it as a constructive conversation opener. And remember that it’s far better than having SEO ignored ☺️ https://t.co/7MyQpHrlBl
Transformer architecture handles language as sequences of tokens. There is no parser inside the model looking for <schema> tags. There is no preference for FAQ markup. The model reads the words. That is the mechanism. 👇
"Ensures AI engines can parse and connect your content."
That's the SaaS pitch for 'Technical GEO.' Schema. Structured data. Clean architecture.
Ensures.
LLMs were built to read unstructured language. There's nothing to ensure.
Read on The Inference: https://t.co/1SoLxa1hn0
Everyone should understand the difference between (1) information retrieval, (2) hypothesis formation and (3) passing a judgement.
LLMs excel at (1), are a great partners for (2) and fail at (3).
Know what you can outsource.
https://t.co/6d3xflITQc via @DuaneForrester, @sejournal
LLM + Search + RAG + Grounding = figures out where the information is actually mentioned, then show the links in the response.
In AIO and AI Mode, you see links attached.
This is only possible because of grounding - it confirms that the information is actually there.
A Google patent describes the process of Google building custom AI-generated landing pages for when people click on search results, instead of going to your website https://t.co/Ak3ULvUdnv hat tip @glenngabe@NarwhalJosh@BrandonLazovic
From Knowledge Graphs to Vector Embeddings, this article explains why content ranks (or gets buried) based on meaning rather than word count.
https://t.co/r85WsjN1JA
Were we wrong about “The Great Decoupling” after all?
In my latest article, I explore the impact of disabling &num=100 along with its impact on rank tracking & Google Search Console data.
Important read. Make sure to check it out + pass it on to others: https://t.co/9mJgWfgIBP
@pedrodias Noticed the same. Also more clicks than ever before not showing the corresponding queries. I think that may be because of the changes in how people are googling - more conversational.