Sign in

Chris Green SEO

@chris-green.net
1.5K followers 650 following 1.8K posts

Technical Director (Torque Partners), experienced search trainer & mentor. #BeOnePercentBetter

PostsRepliesMedia
Chris Green SEO @chris-green.net · 10h
I discuss more as part of the OnCrawl Myths Vs Data Series (link in the comments) www.oncrawl.com/ai/myth-vs-d... 🧵 2/2
oncrawl.com
Myth vs Data: SEO in the AI era - Oncrawl
Six technical SEO experts each take one widely believed myth about SEO and AI visibility and break it with data.
000
Chris Green SEO @chris-green.net · 10h
AI can render content in a browser - it’s pretty good at doing it now. It can also read text in an image and can interact with JavaScript-rendered content. Does that mean AI Chatbots or other search surfaces will do this every time? That’s an assumption that’ll create risks! 🧵 1/2
250
Chris Green SEO @chris-green.net · 29/09/2026
The lack of GSC/BWT for AI search specifically feels quite acute right now, but a cheap (free) probe - even if it isn’t perfect - can help you understand/troubleshoot what issues are stopping your content from being returned by AI search tools. chrisgreenseo.substack.com/p/checking-a... 🧵 2/2
chrisgreenseo.substack.com
Checking if a Page is Part of a Retrieval Pipeline for AI
Without fixed tools to understand indexing/retrieval (Like GSC) what can we do instead.
010
Chris Green SEO @chris-green.net · 29/09/2026
ICYMI - if you want an approach (and tool) to help you understand if your pages are likely part of AI retrieval, then this could be quite useful! 🧵 1/2
chrisgreenseo.substack.com
Checking if a Page is Part of a Retrieval Pipeline for AI
Without fixed tools to understand indexing/retrieval (Like GSC) what can we do instead.
110
Chris Green SEO @chris-green.net · 24/09/2026
Hopefully you never need this tool, but if you’re working on a migration and worry people won’t prioritise or budget correctly, this (or an approach like it) could be well worth your time! migrationrisk.com 🧵 2/2
migrationrisk.com
Migration Risk | Free website traffic risk analysis
Find the pages to prioritise before your website migration. Get a free traffic risk report using your Google Search Console data.
000
Chris Green SEO @chris-green.net · 24/09/2026
Sometimes someone builds a tool I wish existed a long time ago - this time it is Mersudin who has stepped forward. Migration Risk uses your domain and GSC account data to profile the most important pages on your site and the pages that pose the greatest risk if not migrated correctly. 🧵 1/2
120
Chris Green SEO @chris-green.net · 22/09/2026
In this post, I talk about the issue a little more, as well as provide a small tool to help you understand the impacts of your Robots.txt on “AI search” providers in general chrisgreenseo.substack.com/p/when-block... 🧵 2/2
chrisgreenseo.substack.com
When Blocking Bots Becomes 3D Chess
When “block all” or “block none” approaches are potentially risky, what do you need to do?
000
Chris Green SEO @chris-green.net · 22/09/2026
Bot control in 2026 isn’t as easy - it’s not as simple as just using robots.txt anymore. It also needs to involve more of the business than ever, which is both good and bad. 🧵 1/2
chrisgreenseo.substack.com
When Blocking Bots Becomes 3D Chess
When “block all” or “block none” approaches are potentially risky, what do you need to do?
210
Chris Green SEO @chris-green.net · 18/09/2026
The emerging challenge is not simply making content easier for AI to read. It is making meaning harder to lose between retrieval and belief.” Being retrieved isn’t “mission complete”; it isn’t only the first objective duaneforresterdecodes.substack.com/p/ai-search-... 🧵 5/5
duaneforresterdecodes.substack.com
AI Search Didn’t Remove Cognitive Load. It Moved It.
As AI takes over more retrieval and synthesis, consumers do less searching and more verification. That changes what information needs to survive between retrieval and belief.
000
Chris Green SEO @chris-green.net · 18/09/2026
and synthesis process that produced it.” I think there’s a quality issue here, but Duane’s conclusion is more helpful for SEOs and search professionals now: “Our information may be extracted in pieces, combined with other sources, compressed into a smaller answer, and presented… 🧵 4/5
duaneforresterdecodes.substack.com
AI Search Didn’t Remove Cognitive Load. It Moved It.
As AI takes over more retrieval and synthesis, consumers do less searching and more verification. That changes what information needs to survive between retrieval and belief.
100
Chris Green SEO @chris-green.net · 18/09/2026
“Some of [the effort in searching] genuinely went away… That is real value, and it would be strange to pretend otherwise. Judgment and verification become more important when the answer [the AI generates is] already assembled and the consumer did not personally witness the source-selection... 🧵 3/5
duaneforresterdecodes.substack.com
AI Search Didn’t Remove Cognitive Load. It Moved It.
As AI takes over more retrieval and synthesis, consumers do less searching and more verification. That changes what information needs to survive between retrieval and belief.
100
Chris Green SEO @chris-green.net · 18/09/2026
For those who don’t want to research and don’t care about verification, AI search may be a net gain. Those who've been using it for a while may feel that sometimes the process is not much quicker. That’s what I feel at times, at least! Duane does suggest that this shift is better (overall): 🧵 2/5
duaneforresterdecodes.substack.com
AI Search Didn’t Remove Cognitive Load. It Moved It.
As AI takes over more retrieval and synthesis, consumers do less searching and more verification. That changes what information needs to survive between retrieval and belief.
100
Chris Green SEO @chris-green.net · 18/09/2026
When you are searching for something, where would you rather spend the effort: researching or verifying returned results? Duane’s latest Substack explores the idea that AI search hasn't removed the effort needed, but has merely shifted where that effort is required. 🧵 1/5
duaneforresterdecodes.substack.com
AI Search Didn’t Remove Cognitive Load. It Moved It.
As AI takes over more retrieval and synthesis, consumers do less searching and more verification. That changes what information needs to survive between retrieval and belief.
110
Chris Green SEO @chris-green.net · 16/09/2026
This is still a small exploratory test, based on prompts around a single destination, so it should not yet be treated as a general finding. A larger experiment across multiple properties, destinations, intent types and brands would be needed to test whether the same pattern holds consistently 🧵 7/7
010
Chris Green SEO @chris-green.net · 16/09/2026
Brand inclusion (for this brand) is robust to wording changes, while citations are more dependent on retrieval. Changes in prompt wording can alter the search queries(sources) or query fan-outs generated, which can then surface a different set of sources. 🧵 6/7
100
Chris Green SEO @chris-green.net · 16/09/2026
For this brand, prompt rewording had almost no impact on whether the brand appeared in the response, as long as the underlying intent remained the same. Citation behaviour was much less stable. These two outcomes are being driven by different parts of the response process. 🧵 5/7
100
Chris Green SEO @chris-green.net · 16/09/2026
However, when looking at one established brand within the test set, the pattern was notably different: - Brand mentions: ~99% stability - Brand-domain citations: ~35–45% stability 🧵 4/7
100
Chris Green SEO @chris-green.net · 16/09/2026
I compared the brands mentioned and domains cited across the resulting responses. Across all brands and domains (not URLs, intentionally): - Mentions: 62% of baseline brands were retained when prompts were reworded - Citations: 63% of baseline domains were retained when prompts were reworded 🧵 3/7
100
Chris Green SEO @chris-green.net · 16/09/2026
I believe this is one of the larger issues of prompt tracking. I took 32 prompts, rewrote each prompt four different ways while preserving the underlying intent, and tested each version five times (to make a small account for “noise” whilst making it achievable in the timeframe I needed). 🧵 2/7
100
Chris Green SEO @chris-green.net · 16/09/2026
What’s more important: prompt wording or prompt intent? “I don’t think users would search for that” is a common response to outputs of prompt generation or prompt selection. We don’t have a source of data (like AdWords Keyword Planner) to help us adjudicate/validate our prompt selection. 🧵 1/7
240
Chris Green SEO @chris-green.net · 10/09/2026
The overlap of SEO and GEO is not 100%; arguing that is equally as misguided. If you have a separate GEO strategy, you have a governance issue in the making, a fundamental disconnect that is going to cost you. But you have to take the difference seriously. 🧵 3/3
020
Chris Green SEO @chris-green.net · 10/09/2026
Every time someone sees a difference between ChatGPT’s results & “traditional” results, a tier of company/consultant/tool provider steps in to expose what “no one else is talking about” or “SEOs are in denial of” Let’s be honest, that’s someone trying to mislead you to sell more through fear 🧵 2/3
110
Chris Green SEO @chris-green.net · 10/09/2026
This has only gotten truer over time. I love a good Venn, and Cyrus has created a corker. I know I promise to stop re-litigating the SEO vs GEO debate, but it’s refusing to die. 🧵 1/3
160
Chris Green SEO @chris-green.net · 09/09/2026
If we can demonstrate that visual information changes retrieval outcomes, that’s the point where I think a lot more people will sit up and recognise how significant this could be buff.ly/MkYF36s 🧵 6/6
buff.ly
Your images have a new job in AI search
AI doesn't see your images the way a human does. See what it detects, what it misses, and whether the page confirms the right story.
000
Chris Green SEO @chris-green.net · 09/09/2026
There’s a big jump from “this technology exists” to “this signal participates in retrieval” and then another jump to “we should optimise for it.” I think this is one of the most interesting conversations in “AI Search” right now, and I don’t think people are paying enough attention. 🧵 5/6
buff.ly
Your images have a new job in AI search
AI doesn't see your images the way a human does. See what it detects, what it misses, and whether the page confirms the right story.
100
Chris Green SEO @chris-green.net · 09/09/2026
Can the visual content of an image cause a page to be retrieved when the equivalent signal doesn’t exist in the page text? Should it? Yes! Can I win an argument for investment on it? No, not yet. That distinction matters. 🧵 4/6
buff.ly
Your images have a new job in AI search
AI doesn't see your images the way a human does. See what it detects, what it misses, and whether the page confirms the right story.
100
Chris Green SEO @chris-green.net · 09/09/2026
but I think for people to take this approach seriously (because it takes a lot of effort and isn’t something that SEOs were historically that strong in), some experimentation here is crucial. 🧵 3/6
buff.ly
Your images have a new job in AI search
AI doesn't see your images the way a human does. See what it detects, what it misses, and whether the page confirms the right story.
100
Chris Green SEO @chris-green.net · 09/09/2026
- And we have patents describing ways these capabilities could be used. The question I keep coming back to is: how do we measure whether this is actually happening within retrieval pipelines? I’m not saying I don’t believe Myriam - far from it - 🧵 2/6
buff.ly
Your images have a new job in AI search
AI doesn't see your images the way a human does. See what it detects, what it misses, and whether the page confirms the right story.
100
Chris Green SEO @chris-green.net · 09/09/2026
Myriam’s work on the changing role of images in AI search could be huge. But there’s an itch I need to scratch: - We know some models can understand images. - We know multimodal embeddings can put text and images into shared semantic spaces. 🧵 1/6
buff.ly
Your images have a new job in AI search
AI doesn't see your images the way a human does. See what it detects, what it misses, and whether the page confirms the right story.
100
Chris Green SEO @chris-green.net · 08/09/2026
There are many more nuances here to be aware of, but knowing what your data ISN’T will help you present it for what it is useful for. Ultimately, I’d argue that prompt selection is more important than the method of tracking, and that’s a whole different story! 🧵 9/9
010
Chris Green SEO @chris-green.net · 08/09/2026
visibility data from it can be as useful as a market representation; it can still be useful. To be clear, API + Tool calling is not the same as scraping the UI. Scraping the UI is not the same as a “real user” searching. 🧵 8/9
100
Chris Green SEO @chris-green.net · 08/09/2026
and how you choose to convey it. If you use “indicative” rather than “accurate” when describing a UI scrape, it can help people appreciate that someone, somewhere saw this result. For AI tracking, if you treat it as a more abstract method to capture and present data, 🧵 7/9
100
Chris Green SEO @chris-green.net · 08/09/2026
- Wait, “traditional” rank tracking has become less and less accurate, but we (as an industry) don’t seem to be as bothered. Surely the most important thing is how we interpret and present the data? I have come to the view that which method you should use depends on what you need to convey 🧵 6/9
100
Chris Green SEO @chris-green.net · 08/09/2026
- If you call an API tens of thousands of times, you can likely find out the most useful things about a brand/niche from model memory. - But if you scrape the UI 50-100 times and average results, you can get a pretty reliable view on visibility, enough for insights to be made 🧵 5/9
100
Chris Green SEO @chris-green.net · 08/09/2026
- But the API (with tool calling) is nowhere near as close to the UI experience; is it close enough to be valuable? - Personalisation means that UI fidelity is a bit of a fiction - so this can be most misleading 🧵 4/9
100
Chris Green SEO @chris-green.net · 08/09/2026
a more complete view is better where you want to do larger pieces of analysis. My journey here can be summarised as: - Accuracy isn’t possible; LLMs are deterministic and therefore the answers will never be accurate 🧵 3/9
100
Chris Green SEO @chris-green.net · 08/09/2026
- Deep/probing - a more complete view of parametric memory. Usually via the API, calling the “search” tooling if required Initially you’d think the merits of each method were quite easy to understand. More accurate data is better for a low number of prompts, and... 🧵 2/9
101
Chris Green SEO @chris-green.net · 08/09/2026
There are two key schools of thought with AI trackers at the moment - ones that I think will continue to develop further, but are critical to consider. - Fidelity to “real” results - “more accurate”. Usually through scraping the interface of the AI Chatbot 🧵 1/9
210
Chris Green SEO @chris-green.net · 02/09/2026
Some may simply be the product of immature, inconsistent pipelines. The bigger question is: is the difference intentional, useful and strong enough to optimise around? chrisgreenseo.substack.com/p/what-if-th... 🧵 2/2
chrisgreenseo.substack.com
What if the AI/SEO Gap is a Bug & Not a Feature?
Search is changing, but not all these changes are for the better. Over-focusing on these differences may fail to address the long term goals and what ultimately will be successful.
020
Chris Green SEO @chris-green.net · 02/09/2026
What if the SEO/AI search gap is a bug and not a feature? AI systems often surface very different URLs from traditional search. That matters, but we shouldn’t assume every difference reflects a better retrieval model. 🧵 1/2
chrisgreenseo.substack.com
What if the AI/SEO Gap is a Bug & Not a Feature?
Search is changing, but not all these changes are for the better. Over-focusing on these differences may fail to address the long term goals and what ultimately will be successful.
120
Chris Green SEO @chris-green.net · 31/08/2026
Pedro also gives some practical tips as well as some intuitive definitions, but honestly I’m not putting them here; go in and take a look for yourself 😉 visively.com/kb/algorithm... 🧵 5/5
visively.com
How Search Engines Rank Content: Understanding TF-IDF and BM25 | Visively
How search engines use TF-IDF and BM25 to match keywords to content, and why term-based ranking remains essential alongside semantic search.
000
Chris Green SEO @chris-green.net · 31/08/2026
which, at best, may waste your time, and at worst may put your efforts totally in the wrong direction. So describing this BM25 and then RRF (a new(er) concept for many, most likely) together is super, super useful. 🧵 4/5
visively.com
How Search Engines Rank Content: Understanding TF-IDF and BM25 | Visively
How search engines use TF-IDF and BM25 to match keywords to content, and why term-based ranking remains essential alongside semantic search.
100
Chris Green SEO @chris-green.net · 31/08/2026
but there’s a good chance that a primer/fresher could help solidify your knowledge. I’ve said this before, but what I love about Pedro’s writing is how approachable he makes this for search marketers. TF-IDF in particular has been very often misunderstood or misused - 🧵 3/5
visively.com
How Search Engines Rank Content: Understanding TF-IDF and BM25 | Visively
How search engines use TF-IDF and BM25 to match keywords to content, and why term-based ranking remains essential alongside semantic search.
100
Chris Green SEO @chris-green.net · 31/08/2026
But one thing I’ve found over the last year or so of building my own retrieval/text analysis pipelines is that understanding this helps aid your own progress too. You may have heard of TF-IDF and BM25 before - they’ve been discussed for a while - 🧵 2/5
visively.com
How Search Engines Rank Content: Understanding TF-IDF and BM25 | Visively
How search engines use TF-IDF and BM25 to match keywords to content, and why term-based ranking remains essential alongside semantic search.
100
Chris Green SEO @chris-green.net · 31/08/2026
School is in session, and Pedro is taking this class. The core concepts that underpin search ranking (or at least parts of it) are key to understanding at least on a basic level. 🧵 1/5
visively.com
How Search Engines Rank Content: Understanding TF-IDF and BM25 | Visively
How search engines use TF-IDF and BM25 to match keywords to content, and why term-based ranking remains essential alongside semantic search.
120
Chris Green SEO @chris-green.net · 26/08/2026
I’d love to see some bigger, longer-spanning experiments like this, which can help to illustrate wider-reaching conclusions, but the approach IS interesting and worth considering if you haven’t already www.wislr.com/research/do-... 🧵 7/7
wislr.com
Do LLM training pages actually work? Yes, 17% more crawls in a controlled test
Yes. In a controlled eight-week test, the category that got training pages was crawled about 17% more than a comparable one held back as a baseline. One ecommerce product category got prerendered LLM…
010
Chris Green SEO @chris-green.net · 26/08/2026
Not a perfect metric (no one is pretending it is), but a proxy, an early, potentially leading metric that - if studied over months and drawing in other metrics - could eventually link up with other KPIs. 🧵 6/7
wislr.com
Do LLM training pages actually work? Yes, 17% more crawls in a controlled test
Yes. In a controlled eight-week test, the category that got training pages was crawled about 17% more than a comparable one held back as a baseline. One ecommerce product category got prerendered LLM…
100
Chris Green SEO @chris-green.net · 26/08/2026
This study from Tony @ WISLR interests me because they’re testing for training bot hits amongst all else. An experiment to see whether a change in available content provides more cause for a training bot to revisit or not. 🧵 5/7
wislr.com
Do LLM training pages actually work? Yes, 17% more crawls in a controlled test
Yes. In a controlled eight-week test, the category that got training pages was crawled about 17% more than a comparable one held back as a baseline. One ecommerce product category got prerendered LLM…
110
Chris Green SEO @chris-green.net · 26/08/2026
Models take months/years to update, cutoff dates are not always super-clear, and then being confident that a specific change you have made has influenced how a model retrieves information and generates a response is even harder still. 🧵 4/7
wislr.com
Do LLM training pages actually work? Yes, 17% more crawls in a controlled test
Yes. In a controlled eight-week test, the category that got training pages was crawled about 17% more than a comparable one held back as a baseline. One ecommerce product category got prerendered LLM…
100
Chris Green SEO @chris-green.net · 26/08/2026
If we smooth over the bumps and details, it comes to the question of how long optimisation takes to be reportable and how you can report on it. With search sources, it is relatively straightforward: are you receiving linked citations or not? With model memory, it is much trickier. 🧵 3/7
100