At 9:42am on 7 March 2026, I searched Google for “aao optimisation southampton” on my phone.
The first organic result — sitting directly below an AI Overview — was not a service page, a blog post, or a landing page. It was a plain text file: https://www.seostrategy.co.uk/llms.txt. Below it, displaced to second position: the University of Southampton.
I want to be precise about what that means. A plain text file I generated using my own WordPress plugin outranked a Russell Group institution — one of the UK’s leading research universities — on a query containing their own city. Not because of links or domain authority. Because Google crawled, indexed, and ranked a file that was never designed to rank at all.
The Evidence: One Morning, Five Queries
Between 9:42am and 9:54am on 7 March 2026, I ran a series of searches. Here is what each one returned.
“aao optimisation southampton” — 9:42am. Position 1 organic: seostrategy.co.uk/llms.txt. University of Southampton below it. Google’s snippet pulled directly from the file’s content: “SEO Strategy is a specialist SEO & LLM optimisation consultancy based in Southampton, Hampshire.”
“llms.txt southampton” — 9:53am. Position 1: my guide, llms.txt: The Complete Guide to Making Your Site AI-Readable. Clean number one, no meaningful competition.
“llms.txt optimisation southampton” — 9:53am. Positions 1 and 3 both seostrategy.co.uk. The guide at one, the LLM Optimisation service page at three.
“llm optimisation southampton” — 9:54am. The AI Overview for this query did not just link to my site. It named my business: “Specialist consultants like SEO Strategy and local agencies provide services including entity SEO, AI visibility audits, and structured data implementation.” Named entity citation inside an AI Overview is the highest form of AI visibility currently measurable in Google Search.
“aao optimisation specialist southampton” — same session. Position 1: AI Agent Optimisation (AAO): Be Found & Chosen by AI Agents — seostrategy.co.uk again. The site is now the default entity for AI optimisation queries in this geography.
What Google Search Console Confirmed
Checking GSC the same morning: the llms.txt URL showed as indexed — “URL is on Google” — with first impressions appearing in the days immediately before this session. The file had generated 2 impressions at an average position of 48, with the query “perplexity seo consultancy” appearing as its first recorded search term. The spike in impressions visible on the chart is steep and recent — Google indexed and started serving this file in the same week the SERP screenshots were taken.
This is not a long-term SEO play. This happened fast. The file existed, Google found it, indexed it, and started ranking it — all within days.
What the AI Platforms Said
Gemini. Asked “who provides LLM optimisation in Southampton?”, Gemini produced a summary table. Under “Rank in ChatGPT / AI Search”: SEO Strategy Ltd, NRLC.ai. The full response named SEO Strategy Ltd (Sean Mullins) as a highly specialised consultancy based in Southampton, focused on LLM Optimisation services, entity SEO, and structured data. Three sources cited, +2 additional.
Perplexity. Listed SEO Strategy Ltd at position one with full entity detail — consultant named, services described, link cited. The description matched language from the llms.txt file and the site’s structured content.
ChatOn (GPT-based). No mention of SEO Strategy. Returned a generic response listing universities, research institutions, and unnamed tech companies. This is expected — different AI systems have different training cutoffs and citation architectures. It also demonstrates that citation dominance in one platform does not automatically transfer to all. The work continues.
The Reboot Study — What It Actually Tested
In February 2026, Reboot Online published a three-month controlled experiment concluding that AI bots do not visit llms.txt files. The study was methodologically careful: they published pages with zero internal or external links, listed them only in llms.txt, and monitored log files for AI bot activity. None arrived.
That finding is real and worth knowing. It is also a test of a specific and narrow claim: whether llms.txt helps AI bots discover orphan pages with no other links. Almost no real implementation works that way. Real llms.txt files link to existing, already-indexed pages. The question practitioners actually care about is different: does Google index the file as content, does it contribute to entity signals, does it influence AI Overview citations?
The screenshots above answer all three. Google indexed the file. The entity signals contributed to a named citation. The file itself ranked in SERPs. Reboot’s study and this evidence are not in conflict — they are answering different questions. The newsletter summary that called it “AI bots give llms.txt the cold shoulder” was accurate about the specific test and misleading about the broader picture.
The Mueller Question
In late 2024, Google’s John Mueller compared llms.txt to meta keywords — a well-intentioned file that search engines would likely ignore. The comparison has been widely repeated as a reason not to bother. With respect to Mueller, whose guidance has consistently been valuable, empirical evidence has to take precedence over a theoretical assessment made before the evidence existed.
Google indexed this file. Google ranked it. Google used the signals within it to produce a named entity citation in an AI Overview. That is not the behaviour of a file being ignored. Mueller’s assessment may prove correct at scale as adoption widens. Right now, in March 2026, first movers have verifiable advantages that cannot be dismissed as noise.
The Local SEO Dimension
Every result in this session carried a Southampton geographic anchor. The llms.txt file mentions Southampton, Hampshire, and London naturally throughout. It links to pages that carry the same topical and geographic signals. When Google’s AI systems processed the file, they had everything needed to build a clear, location-anchored entity model.
The result: AI Overview citation for a local commercial query. Named. Linked. Authoritative. For local businesses — solicitors, tradespeople, clinics, consultants — this is significant. The businesses that give AI systems a structured, location-specific signal about who they are and what they do will be cited. The ones that don’t will be invisible to a growing share of discovery behaviour.
I have a client site — a dog walking business in Portsmouth — that has ranked #1 in Google for seventeen years on clean code and good content. That site has no llms.txt. A competitor with a weaker traditional SEO profile but a well-structured llms.txt may already be building AI citation dominance that will matter more than those seventeen years of rankings within two years.
The Commercial Case
Generating a llms.txt file using the LLMs.txt Curator WordPress plugin takes under five minutes. The file that produced these results cost effectively nothing to implement. The upside includes: additional rankable real estate you didn’t know you had, AI Overview citation, entity signal reinforcement, and named brand mention across Gemini and Perplexity for high-intent commercial queries.
For context: I work with B2B clients at £600–£900 per day. One inbound enquiry driven by AI visibility recovers implementation cost immediately. The question is not whether llms.txt is worth doing. The question is why it hasn’t been done yet.
A Note on Indexing Control
Not every brand will want their llms.txt appearing in SERPs. A plain text file is not a polished landing page. For enterprise clients where brand consistency matters, preventing the file from surfacing in results makes sense. The upcoming update to the LLMs.txt Curator plugin will include an X-Robots-Tag noindex toggle — off by default (preserving the ranking benefit) but available for brands who prefer to keep the file serving its AI-navigation function without appearing in Google results.
What to Do Today
Implement llms.txt immediately. If you’re on WordPress, use the LLMs.txt Curator plugin. Write your descriptions carefully — the language you use in the file is what Google will pull into search snippets. Treat it as a brand document, not a technical artefact.
Submit to Bing Webmaster Tools. Microsoft Copilot draws from Bing’s index. One submission could produce Copilot citations within days for fresh, specific content in low-competition categories.
Check GSC for the URL. Once your file is live, go to Google Search Console and inspect the /llms.txt URL. If it’s not yet indexed, request indexing. Once it appears, filter GSC performance by that URL to see which queries are surfacing it. Those queries are a window into how AI systems are categorising your entity.
Think of it as a second About page — one written for machines, indexed by Google, and read by every AI system that touches your domain. The difference is that this one can rank.
All screenshots referenced in this post were taken on 7 March 2026 and are available in the original published version of this article.
