🖥️

​Desktop accounts for over 90% of referral traffic from AI search. This may be skewed because Apple's Safari, is the default browser for nearly 1 billion users and doesn't have AI search included by default...yet.

🫥

​Google Search Console fails to report on 50% of queries. This is particularly apparent with conversational-style/long-tail queries, which are likely to grow in usage in the near future due to Google's AI Mode.

📉

​GSC Discover report bug confirmed by Google. Some users shared that their performance reports dropping to zero clicks & impressions for Google Discover and Google has since added an annotation to 5 June 2025.

📅

​The average #1 ranking page is 5 years old. This has increased from two years old in 2017. This study also found only 1.74% of newly published pages rank within the top 10 within one year.

▶️

​AI Overviews and Copilot Answers include short videos in results. These results include both YouTube Shorts and TikTok videos, reinforcing the importance of a multimedia, multichannel approach to content.

🔍

​Perplexity CEO confirms 780 million monthly queries and growing. This may seem like a large figure at first but compared to Google's 98 billion searches per week, Perplexity is still fighting for its market share.

📊

​Google confirms visualisations included in AI Mode. As announced at Google IO last month, AI Mode can now create its own tables, graphs and charts, beginning with financial and sports queries.

🛑

​Google provides guidance if your site is incorrectly flagged as explicit. The guidance includes using site: operator combined with SafeSearch, updating affected pages and requesting a review.

🍳

​Google updates guidance for images in recipe structured data. Google highlights the importance of sharp, high-quality images and how this can directly the ranking of images in search.

Oh no!

With AI Mode, how will it be possible to find the next most likely intent search from a super longtail query in their search fan out?!?

Don't forget what humans like

The idea of improving content using embeddings feels somewhat dissonant to the “create content for humans, not for robots” that many SEOs (admittedly, not all) have been championing for years. 🤔

I said on stage last year at SERP Conf. Vienna that I believe much of the future of “SEO” would be about robots discovering things and fetching them for humans. However, it’s important to remember these robots are being trained to find what humans like.

Optimising for a metric this way means you are converging on a historic data point, as intent shifts, culture shifts, new memes are created, and news happens, focusing on a numerical representation of the past is not going to be the key to big wins in the future.

It will kill creativity.

I saw Tinder’s #MoveOn “Ex-press Disposal Truck” campaign today via Phil Blondé. This kind of creativity will drive press headlines, links, branded search, and references all over the web on ‘relationship sites’ for months to come.

The breadcrumbs of LLMs lead towards the mean; and that’s where you never want to be in marketing ❤️

Good job, Tinder team!

Original doesn't always mean unique

In my experience, people confuse "original" with "unique". At Aleyda's SEOFOMO event in London, Gary Illyes said Google would be prioritising "original" content in the future. 🧝‍♂️
​
If you use an LLM to generate your content, you will get something that is unique, but not original.
​
If you take your own data, stories, experience, and run it through an LLM to write, it will be original, but even then you'll be losing one "edge" of originality, which is the ability to intelligently apply style.
​
I think we'll collectively develop a new type of "ad blindness" to GPT-style writing. I've already caught myself starting to read things, immediately sensing they are GPT-written and moving on.
​
Original is going to be become even more scarce and valuable.

Google Query Parameters

Here is what I believe to be a complete list of the 674 query parameters accepted by Google dug up from source code. 📃
​
Why? Because having some exposure and understanding how systems work can give you "eureka!" moments about the bigger picture when you are making your own observations about outputs.
​
I have annotated some of the more obvious ones, but if you recognise a query parameter that I haven't listed, please email me so I can update and credit you 😎
​
​Google Query Parameters​

Trailing slashes; still forgotten in 2025

If your web page URLs work both with and without a trailing slash (/), search engines will view these as two separate and competing pages. You should pick whether you want a trailing slash or not and setup permanent 301 redirects between each and internally link consistently to the correct one. Failing to do so, can result in the 'two' pages competing against each other in the SERPs and ranking worse than a single one.

One declaration to rule them all

Make sure you're only specifying hreflang with one method (on-page, sitemap, headers). I've seen numerous problems caused with conflicting tags - so check you're only using the one method!

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)