💰

​Ads now appear in nearly 30% of AI Mode queries. Interestingly, CPC appears to be the strongest determining factor, with ad presence increasing to 53.56% with a CPC of $10 or higher.

🥈

​Google is now the second-most cited domain in AI Mode. This increase in the last few months is primarily driven by Google Business Profiles in local searches and Product Knowledge Panels in comparison searches.

🤝

​AI Mode now connects to Canva, YouTube Music and Instacart. This Connected Apps feature is different to Personal Intelligence; this allows users to control these external partners directly through AI Mode.

🧑‍⚖️

​Google must share search data with competitors and integrate competitors' AI equally on Android devices. These are the two declarations from the European Commission as part of the Digital Markets Act.

📊

​Google's Core Web Vitals baseline LCP may not be accurate. A study found that optimal LCP ranged from 100ms to 1s, rather than the typical 2.5s recommendation. The author clarifies that this is NOT a new target to chase.

📰

​Using an "Are you a robot?" page can cause indexing issues for Google. John Mueller explained that these pages can be indexed and treated as a duplicate of the page the crawler was originally trying to access.

🫕

​NotebookLM rebrands to Gemini Notebook, including changing its user-agent. Webmasters will need to update their .htaccess or firewall to block this new user-agent as these fetchers don't obey directives such as robots.txt.

We analysed 10 billion web requests across WordPress sites.

AI bot traffic grew 300% in a single year. By late 2025, one in every 31 web requests came from an AI crawler. Scrapers, broken automation loops, and aggressive crawlers are hitting WordPress sites harder than most people realise.

Kinsta analysed more than 10 billion requests to understand what is actually happening, how modern crawlers behave, and what site owners should do about it.

If you like nuanced answers, follow my Substack

If you're subscribing to Core Updates, you're probably like me, and like your information quickly, and in bite-sized chunks. However, sometimes this can lead to misunderstandings, and some topics deserve more exploration.

That's why I've started a Substack where I take deep-dives into these topics. Last week, I wrote about:

• ​The AI Convergence Problem: Where LLMs are weak, they're stupid. Where they're strong, they're dragging your marketing toward the mean.
• ​Schema, LLMs and the Low Bar for “Evidence” in GEO: I built a fake company with nonsense schema. The LLMs returned the address anyway. That is not the win the GEO industry thinks it is.
• ​ChatGPT Is Secretly Googling Things. QueryFan.com Shows You Exactly What: ChatGPT and Gemini don't just answer questions. They fire traditional searches in the background, and the results of those searches determine whose content gets surfaced.

It would be great to have you along for the ride! Anyway, onto this week's tips.

Rocky watch whole SEO die. Could not fix.

If your URL is cited in an AI Overview, and the same URL is listed in the classic '10 blue links', is that 1 or 2 impressions in Google Search Console? ⤵️
​
The answer is 1*. I thought I would share it, as it was not obvious to me, so I am sure to at least one other person out there, the answer is not obvious either!
​
​Jamie Indigo went on the hunt for the answer to this after we spoke about it earlier in the year.
​
*Assuming GSC is not showing random numbers for impressions, which does happen sometimes 🎲 🤡

Agents. Agents. Agents.

We are in a sea of new proposals for standards, protocols, and acronyms, with everything ranging from website functionality, to website discoverability. 🫨
​
Luckily for you, Chris Green has pulled together this excellent visualisation to help you mentally map how mature these standards are, whether they are for knowledge or functionality and whether they are website or agent-controlled.
​
Enjoy!

How do you know when the AI is done?

After candidates are retrieved, AI systems are more likely to return/cite content that looks like it was written by AI systems*.

Do with that information what you wish 🧙‍♂️

*This is not because they were written by AI, but their 'expectations' are formed by their own training data, which is expressed in their output.

Mark and Jack are two distinct individuals

Since 5.4, ChatGPT query fan outs are doing specific site: searches. It is vital information to find out where you need to be to be. I spoke about this with Vince Nero a few months ago on the BuzzStream podcast.
​
You can use QueryFan.com to find these at scale 😎

Gemma4 is the bravest LLM I have ever met.

There are some really capable LLMs like Gemma4 that you can run locally to perform tasks that would be otherwise prohibitive because of token cost.
​
A great example of this at Candour is how we're massively speeding up some tasks such as topic/sub-topic categorisation and gap analysis with in-house tools that can run on entire pages of content grabbed with Screaming Frog.
​
Dream time to work in SEO where you can have the data served up and use time to actually engage your brain on how to win 🧠

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)