Ben Rees

I am invisible on every AI platform. Here is the data.

For ten weeks I ran an AI visibility measurement system across ChatGPT, Gemini, Copilot, and Perplexity. My appearance rate for non-branded queries is zero. Here is what the data actually shows.

Ben Rees - 29 June 2026

For ten weeks I have been running an automated agent that fires queries across Google, ChatGPT, Gemini, Copilot, and Perplexity to measure whether bjrees.com appears in AI-generated answers. The queries include "how to measure AI search visibility," "how to get cited in AI-generated answers," and "what is the future of SEO in the age of generative AI." These are the topics I write about and the space I am positioning myself in.

Across ten weeks, 78 runs, and every AI platform tested, my appearance rate for non-branded queries is zero.

Not low. Zero.

The measurement

The system, which I call Klara, separates queries into three types. Branded queries ("site:bjrees.com," "bjrees generative engine optimization") are system health checks. They confirm the crawler is working, not that the content strategy is. What matters are the non-branded queries: searches where bjrees.com would have to earn its place in the answer rather than be looked up by name.

For the AI Search Visibility category, the results across all platforms over ten weeks:

  • ChatGPT: 0 appearances
  • Copilot: 0 appearances in 12 queries tested
  • Gemini: 0 appearances in 12 queries tested
  • Perplexity: 0 appearances in 46 queries tested

This is not a run-by-run anomaly. The pattern is consistent from the first capture in April to the most recent one this morning.

Why the obvious explanation is wrong

The intuitive hypothesis is that the content exists but AI platforms are ignoring it. That would be a citation architecture problem: the right content in the wrong format, or with insufficient external corroboration to be retrieved confidently.

It is not that.

I pulled Google Search Console data for bjrees.com this week. Of the twelve queries Klara tracks in the AI Search Visibility category, eleven have zero GSC impressions. Not a low ranking. No ranking. Google has no record of bjrees.com appearing for "how to improve AI visibility for a B2B website" or "how to get cited in AI-generated answers."

AI platforms retrieve from what they know. If a page does not exist in Google's index for a query, it does not exist for Perplexity either. There is no citation architecture to optimise. There is simply no content.

The one foothold that exists

The only non-branded query cluster where bjrees.com has any Google presence is SMB-to-enterprise scaling: "scaling marketing from SMB to enterprise," position 1.3, 95 impressions over 90 days, 2 clicks. One thin cluster, performing modestly, in a category I am not primarily trying to own.

That ranking also dropped once between captures this week. A cluster of five pages is not topical authority. It is a coincidence that can be reversed.

What the GSC data also found

The top query by impressions for bjrees.com over the last 90 days is "kings cross to green park." 575 impressions. 2 clicks.

A 2021 post using the London Underground as a metaphor for achieving objectives had been ranking, apparently for years, as a tube journey planner. It generated more search impressions than every piece of positioning content on the site combined.

That post was deleted this morning. It is a fair summary of where things stand: accidentally visible for things that do not matter, invisible for things that do.

What fixes this

Both gaps have the same root cause: the content does not exist. No amount of structured data, internal linking, or citation seeding changes that.

The priority order is straightforward. Write content that directly answers the queries being tracked. Get those pages indexed by Google. Then work on the citation architecture that determines whether AI platforms retrieve them. That sequence cannot be reversed.

The SMB-to-enterprise cluster needs deepening, not abandoning, because it is the only compounding asset on the site. Pivoting to GEO content before consolidating what already works is how you end up with zero traction in two categories instead of one.

Why this is worth publishing

Klara is uncomfortable to run when the findings are this clear. It is also the only honest way to know where you actually are rather than where you assume you are.

Most people building content in the GEO/AEO space are optimising content that already exists for queries that already have search volume. That is not where I am. I have one working cluster, eleven invisible queries, and a measurement system precise enough to tell me exactly which problem to solve first.

The score is zero. The work starts here.


Klara tracks bjrees.com visibility across Google, Google AI Mode, ChatGPT, Gemini, Copilot, and Perplexity. It runs overnight and emails a status each morning. The data here covers 78 capture runs from April to June 2026.