✍️

​Google confirms that AI-generated titles are being tested in search results. Of course, Google rewriting titles is nothing new but this is the first time AI-generated titles have been confirmed.

🫸

​Google confirms development of controls to opt out of generative AI features. This is in response to the Competition and Markets Authority's proposals to keep search results fair and useful.

🛒

​Google announces new capabilities for Universal Commerce Protocol. These new features include a new cart option for purchasing multiple items, a new catalogue for product details and a way to link to users' loyalty benefits.

❇️

​Google's Personal Intelligence expands to AI Mode, the Gemini app and Gemini in Chrome for US users. Personal Intelligence can make suggestions based on your activity in connected apps such as Gmail and Google Photos.

📰

​Breaking news traffic has increased by 103%, despite 42% drop in clicks for publishers. Google Discover and the Top Stories section in Google News have been a particularly good source of traffic for some publishers.

🥇

​Google dominates AI search in terms of usage, cost-effectiveness and integration. Despite ChatGPT's early momentum, and Google seemingly caught unprepared, Google seems to be reclaiming its top spot.

🛍️

​New rules in Google Merchant Center require buy buttons for out-of-stock items to remain visible. The buy button for out-of-stock items can no longer be hidden and must be visibly disabled and greyed out.

LinkedIn is an AI citation machine. Take advantage of it.

LinkedIn ranks #2 for AI citations across ChatGPT, Perplexity, and Google AI Mode, appearing in 11% of AI responses.

With semantic similarity scores of 0.57–0.60, AI models are closely mirroring your LinkedIn content in their answers. Are you taking advantage?

Enterprise AI Optimization lets you track and steer how your brand shows up in AI search, across your markets and domains.

See what Semrush sees and stay ahead.

'More Google than Google' is our motto

Google Discover is a huge source of traffic that is commonly overlooked, I thought I would ask Google Discover expert Clara Soteras Acosta for some advice on how she generates traffic, she said:
​
"For me, it’s vital to choose a strong headline and a powerful image. Those are the two main triggers for getting a click in the channel feed. The image needs to be striking enough to stand out from the competition when someone is scrolling through the feed, and the headline should hold something back and create a curiosity gap. The goal is to encourage the click without turning it into clickbait—just writing the headline creatively without giving away the answer in either the image or the headline."

Wake up, time to work

"Can we start with a lower SEO budget and increase when we see results?" - Here is an analogy you can use with your boss: ⬇
​
Starting SEO is like joining a marathon after it has started. You are competing with people that are ahead of you.
​
Does "I am going to jog slowly, and if I start to catch up with them, then I will run harder to try and win" sound like a reasonable plan? Of course not. It will simply end with you expending some energy for no gain.
​
The same is true in SEO. If you want to catch up, then you have to face reality and realise people are not there by chance and you need to outwork (or outsmart) the competition, which usually requires more effort, not less. 💅
​
As I tell clients, "jump with both feet, or don't jump".

I've seen queries you people wouldn't believe

The Google exploit we found showed Google classifies almost all queries into one of eight 'refined query semantic classes', they are:
​
🌟 Short fact
🌟 Bool (short for Boolean – yes/no questions)
🌟 Other
🌟 Instruction
🌟 Definition
🌟 Reason
🌟 Comparison
🌟 Consequence
​
I fine-tuned DistilBERT model trained on 4.8M queries so you can see how Google classifies queries; and made it a free tool for you. 🤝
​
​Here's my talk from brightonSEO last April where I explained this in more detail.​

How can an LLM not know what it is?

Ever seen someone ask an LLM how it ranks document and expect it to answer as if it has knowledge of its own internal workings? 🤦‍♂️

LLMs are not ‘self aware’ in that they are accessing their own code, systems, ranking, preferences and answering your questions about them. Outside of their system prompts, they aren’t going to tell you how they generate answers, or what they are looking for or how you can optimise for them.

When you ask LLMs these kind of questions about themselves, or in fact, any type of system, you’re simply getting back what other randos on the internet have written which has been absorbed in training data, or fetched with grounding.

Unfortunately, this leads to a viscious confirmation cycle of people asking LLMs things, thinking it is correct, writing another article saying this, which then gets regurgitated by the endless LLM centipede. 🤡

What people write may or may not be correct, but don’t think the LLM is going through its own code base and advising you as such 🤓

I can't rely on my keywords

Heard of "The Elliot Effect"?

​Dorron Shapow says modern marketers are too focused on optimising for algorithms and keywords rather than human psychology and has written an amazing guide on how to effectively use AlsoAsked data.

Top Core Updates referrer leaderboard

A big thank you to our top referrers, who have signed up over 10 people to the Core Updates newsletter, go follow them!

🥇 MJ Cachón Yáñez (LinkedIn / Bluesky)

🏅 Nikki Pilkington (LinkedIn / Bluesky)

🏅 Lidia Infante (LinkedIn / Bluesky)

🏅 Saijo George (LinkedIn / Bluesky)

🏅Alice Rowan (LinkedIn)

🏅Martijn Jeurissen (LinkedIn)

🏅Chester Beard (LinkedIn)