Blog
-
The AI report in Search Console shows impressions, not clicks. Here is how I read it
Search Console’s Generative AI performance report has been live for every site since 31 August 2026. It shows impressions in AI Overviews and AI Mode by page, country, device and date. It shows no clicks and no queries. Read it beside the ordinary Performance report, page by page, and you have better data than any…
-
Blocking Google-Extended will not take you out of AI Overviews
Blocking Google-Extended stops Gemini training and some grounding outside Search. Google says it does not affect inclusion or ranking in Search, so it will not take you out of AI Overviews. Those follow the ordinary snippet and index controls, which also change your normal results. For most local and ecommerce businesses, I would block nothing.
-
Schema will not get you into AI answers. Here is what it is still for
Google says structured data isn’t required for its AI features and there is no special markup to add. Schema still earns rich results where Google offers them, and that list keeps shrinking. Mark up what is true and visible, for the results it unlocks. Most of the markup I check is broken anyway.
-
Query fan-out is how Google searches, not a list of pages to write
Query fan-out is how Google’s AI features search: many related queries behind one answer. It is not a list of pages for you to write. When I sized one vertical on this site, 158 keyword rows turned out to be 34 problems written four ways. A page per variation is what Google calls scaled content…
-
AI Overviews run on the index you already have. Read a GEO proposal with that in mind
Google says its AI Overviews and AI Mode are built on core Search ranking and pull from the same index, with no extra requirements to appear. Most GEO proposals I am shown sell the things Google’s own guide says are unnecessary: llms.txt, chunked content, special schema, a page per question, bought mentions. Read the guide…
-
Google lists sixteen manual actions. A local business can realistically get about five
Google lists sixteen kinds of manual action. A local business or a store can realistically get about five of them, and most of those come from something done years ago by somebody else: bought links, near-identical city pages, review markup on your own reviews, comment spam, a keyword-stuffed footer. One more can happen without you…
-
Hidden text breaks Google’s spam policies. Tabs and accordions don’t
Google’s hidden text policy is about intent: content placed solely to manipulate search engines. Accordions, tabs and screen-reader text are named as fine. The hidden content I actually find is accidental, like headings bound to a logo or empty headings from a page builder. That is not spam. It is still a wasted slot, and…
-
Penguin went into core ranking in 2016. The lift from bought links does not come back
Penguin stopped being an update in 2016 and now discounts spam links instead of demoting whole sites. Google’s own documentation says the ranking benefit of neutralised links cannot be regained. So a links-driven slide is usually the end of a lift that was never secure. The work is earning its replacement, not recovering it.
-
Panda became part of core ranking in 2015. Most of the Panda problems I find now are templates
Panda has not been a separate update since 2015, and the helpful content system joined it inside core ranking in March 2024. There is no refresh to wait for and no page score to fix. On the catalogues I audit, half the index being thin is normal, and the fix is a template decision, not…
-
When it really was Google, the next question is which part of Google
Most drops I’m called in for are not Google. When the tells say it was, the name people use is usually years out of date: Panda and Penguin stopped being separate updates in 2015 and 2016. What matters is which system hit you, because a core update, a spam system and a manual action each…
