A week ago a competitor's ranking page pushed us from 1st to 6th in our own UK B2B Tech PR AI Visibility Index, and we responded by rebuilding the agency list on our category page. We're still 6th. Clarity has moved to the top of the table this week, and most of that change comes from one assistant.
There's better news underneath the headline. Other sources are mentioning us more each week, which is the part of AI visibility that is hardest to take away.
What happened in week 4 of the index?
Every week we ask AI assistants the same 86 questions a buyer might ask when looking for a UK B2B tech PR agency, and we count which agencies they recommend. Clarity came first this week with 8.8% of the weighted share of mentions. Hotwire followed on 8.5%, then Brands2Life on 7.5%, Wildfire on 7.3% and Archetype on 6.0%. We were 6th on 5.4%, the same place as last week.
Five assistants count this week. Google AI Mode didn't return its answers because of a problem at our data supplier, so it's missing from the table, as AI Overviews was in week 1.
Why did Clarity move to the top?
Most of it happened on ChatGPT, which carries the most weight in the index. ChatGPT named Clarity in 49 of its 83 answers this week, up from 37. Its single most-cited source was Clarity's page listing the UK B2B tech PR agencies it rates most highly, the same page that moved the table last week.
The other assistants used that page less. Across all five it was cited in 74 answers, down from 90 on the same five last week. Clarity's lead comes from one assistant reading one page very closely, and because ChatGPT carries more weight than any other assistant in the index, that one close reading is enough to put it at the top of the combined table.
ChatGPT's answers also changed shape this week. Almost all of them came back as numbered lists, and each one cited just two sources, where a typical answer in week 3 cited five or six. With only two places behind each answer, the page that takes one of them counts for much more, and Clarity's list took one in 32 answers.
What ChatGPT recommends held steady, though. Each week we ask it a set of check questions that we don't score, to see whether its answers stay the same when nothing else has changed, and this week they did. We're looking into whether the new format comes from ChatGPT itself or from the way its answers reach us, and we'll report what we find.
What did our own changes do?
Last week we rebuilt the agency list on our B2B tech PR page as a comparison table, which covers 13 UK agencies in the order of this index and updates itself every week.
The assistants are reading it. Our site was cited in 188 answers this week and that page alone in 116, which makes it the most-cited page in the study again. ChatGPT named us in 37 of its answers, down from 47, while Perplexity named us in 31, up from 27. Grok still puts us first.
Our next step is to try to win back ChatGPT's preference. Until today the table sat below a section about us, so a buyer asking for the best agencies met our pitch before the answer. We've now moved the comparison to the top of the page, so the answer comes first. Next week's run will be the first to see the change, and we'll report whether it helped.
Where is the good news?
When we leave out answers that were reading our own website, we rose from 16th to 10th. It's our third rise in a row on that measure, after 58th in week 2 and 16th in week 3.
This is the number we care about most. A position that depends on our own pages can be overtaken by one competitor page in a week, as we've now seen twice, whereas mentions in other people's lists and articles take longer to win and hold once they're there.
The work still to do: earned media and offsite citations
When buyers ask for agency recommendations, the assistants cite a long tail of third-party lists. The lists from Startup Magazine and Silicon Canals were each cited in more than 30 answers this week, and PRWeek's 2025 technology table in 28. Neither of the first two mentions Resonance. Clarity barely appears on them either, so they're open ground for every agency in the table.
That's where our effort goes next. We'll work to earn a place on the lists and directories the assistants already read, and we'll pitch the story of these four weeks to the trade titles that cover our industry. Our research on earned media in the age of AI answers looks at which publications the assistants cite when buyers are choosing a supplier.
What can other B2B companies take from this?
- Look at each assistant on its own. One assistant can move a whole ranking, so a blended score can hide what's really going on.
- Put the answer first. If a page answers the buyer's question, open with that answer and keep your pitch for further down.
- Track your offsite visibility separately. The view without your own website shows whether other sources are starting to mention you. It moves more slowly than the headline, and more reliably.
- Expect the assistants to change too. A fixed set of check questions, asked every week, tells you how much of a shift is down to you and how much is the assistant.
The full ranking and every week so far are on the index page. If you'd like to know which sources the assistants read in your own category, we can run a free AI visibility snapshot for you.
.png)