This week’s stories centre again around the fallout of Google’s removal of num=100
Google’s Wild August 2025 Spam Update Finished Rolling Out
https://www.seroundtable.com/google-august-2025-spam-update-done-40151.html
Google’s August 2025 Spam Update ran from 26 August to 22 September, lasting 27 days. It was a broad global update targeting sites that violated spam policies, not specifically link spam. Many websites saw changes within 24 hours, with volatility continuing into September. Some sites lost visibility, while others that were previously penalised recovered. The rollout also coincided with Google’s change to how the num parameter works, affecting scraping tools and impression counts in Search Console. Sites hit are advised to review spam policies and note that recovery may take months.
77% of sites lost keyword visibility after Google removed num=100: Data
https://searchengineland.com/google-num100-impact-data-462231
Google’s removal of the num=100 parameter has caused notable drops in Search Console reporting, with 87.7 per cent of sites seeing reduced impressions and 77.6 per cent losing unique ranking keywords. Short- and mid-tail queries were most affected, while results beyond page three are now far less visible. At the same time, more keywords are clustering in the top three and on page one, indicating the data is now cleaner and less distorted by scraping activity.
For SEOs, this marks a shift in how to read and act on performance data. Impressions and average position are less reliable as benchmarks, since past figures were likely inflated by bots and scraping. The focus should move to meaningful outcomes such as clicks, conversions and branded visibility, alongside rankings that directly contribute to business value.
Schema and AI Overviews: Does structured data improve visibility?
https://searchengineland.com/schema-ai-overviews-structured-data-visibility-462353
A controlled experiment tested whether wa ell-implemented schema (structured data) gives a content page an edge in Google’s AI Overviews. Three near identical pages were created: one with strong schema, one with poor schema, and one with none. Only the page with a well-structured schema showed up in an AI Overview and also ranked highest in traditional search results. The others either failed to appear or performed worse.
These results suggest that the quality of schema, not just its presence, may influence whether Google’s AI cites a page. For marketers, this highlights the importance of implementing meticulous, standards-compliant structured data, especially if you aim to gain visibility in AI-driven responses.
LLMs.txt For AI SEO: Is It A Boost Or A Waste Of Time?
https://www.searchenginejournal.com/llms-txt-for-ai-seo/556576
Many tools and plugins are promoting LLMs.txt as a way to boost visibility in AI-driven search, but it’s purely theoretical; no AI platform currently uses it. Google’s John Mueller confirmed that it is unnecessary, comparing it to the obsolete keywords meta tag. Instead of relying on an unused file, SEO practitioners should focus on strong on-page content, structured data, and clean signals that AI systems can crawl naturally.
The hype around LLMs.txt tends to feed itself: tool makers add support because users demand it, which reinforces the belief it’s needed, even though it has no proven benefit. Rather than chasing speculative optimisations, marketers should invest their energy in practices that already influence AI and search systems in the wild.
The origins of SEO and what they mean for GEO and AIO
https://searchengineland.com/origins-seo-geo-aio-462480
The article explores the evolution from classic SEO to emerging concepts like GEO (Generative Engine Optimisation) and AIO (Artificial Intelligence Optimisation), arguing that regardless of the acronym chosen, “optimisation” remains central. It traces how “SEO” was coined in the mid-1990s by a handful of practitioners such as Bruce Clay, Bob Heyman, John Audette and Viktor Grant, who independently settled on a term to describe the process of improving a site’s visibility. The article points out that just as debates swirled over naming “search engine optimisation” back then, today’s community is wrestling with whether GEO, AIO or another label best captures the evolving landscape.
There’s a grammatical quirk underlying all these terms: can you really “optimise” an engine? The author reminds us that the phrase has always meant “optimising for” an engine rather than optimising the engine itself. He suggests that GEO and AIO risk repeating that same ambiguity. The key takeaway is that while terminology is up for debate, the goal remains consistent across eras: adapting content and signals so that new systems, whether AI engines or traditional search, can understand, trust and surface it.
Also been reading and watching…
Google AI Mode Agentic Features Go Live For Some
https://www.seroundtable.com/google-ai-mode-agentic-features-live-40163.html