A week ago we asked five AI assistants 72 questions that buyers ask about PR software, and counted which tools they named. Agentcy came 74th of 77, named in one answer out of 357. We published that result on our PR software AI index and said we'd run it every week, whatever it showed.
The second run finished last night. Agentcy is now 26th of 76, named in 14 of 349 answers. Here's what moved, what didn't, and what we think caused it.
What changed in between
One thing: our rebuilt website went live on 17 October, three days before the second run. It added guides to free PR software, PR software costs and PR measurement; comparison pages for the best PR software, the best AI visibility tools and the usual alternatives searches; and a pricing page with every plan's price and limits written out. The questions, the assistants, the location and the method stayed the same.
Where Agentcy is now named
Eight of the 14 answers recommend Agentcy. They came from five questions:
- Which PR software has a free plan? Perplexity and Google AI Overviews both list Agentcy, quoting the free plan's limits accurately.
- Is there free PR software with a journalist database? ChatGPT puts Agentcy first, as the best all-round free option.
- Which PR tools can I sign up for myself without a sales call? Perplexity names Agentcy's free plan.
- Which PR platforms let agency clients log in to see their own coverage? ChatGPT lists Agentcy third.
- Which PR software is priced per client for agencies? Grok names Agentcy and cites our pricing page, as it did in the baseline, the one answer we had then.
Perplexity and Google AI Overviews also name Agentcy when asked which PR software UK B2B tech PR agencies use.
Every one of these is a question with a specific, checkable answer, and every answer repeats a fact that is written plainly on one of our pages: a free plan's limits, a per-client price, client logins. None of them is a broad question like "what is the best PR software?"
Read, but not yet recommended
The other six answers are all ChatGPT, and they're a different thing. ChatGPT used one of our guides as a source, so "Agentcy PR" appears as a citation, but the tools it recommends are other people's. That happened on questions about what PR software costs, which vendors publish their prices, the best UK journalist database, and the best PR software for a small agency and for a one-person comms team.
Our scoring counts these, because the brand appears in the answer. But they're worth separating. Being cited as a source means the assistant read your page and trusted it enough to quote it. Being recommended means it believed you belong on the list. The first usually comes before the second.
Across all assistants, agentcy.com.pr was cited in 11 answers this week, up from one. The free PR software guide was cited most, in four answers, followed by the best AI visibility tools comparison in two. No answer cited the index page itself, so it isn't inflating our own number.
Where we should be named and aren't
The gaps say more than the gains. The question closest to what Agentcy does is "Which PR tool tracks AI visibility as well as media coverage?" PRCoverage was named in all five answers; Agentcy in none. The same goes for "How can a PR agency offer AI visibility tracking to its clients?", "Which AI visibility tools have agency plans or per-client pricing?", and "What tools do PR agencies use to manage media monitoring, pitching and reporting in one place?"
The three alternatives searches (Cision, Meltwater and Muck Rack alternatives) went to the large platforms in every answer. We have a page for each, but they're three days old, and assistants with web search lean on pages that search engines already rank. We'd expect those to take weeks, not days.
The rest of the table
Muck Rack overtook Cision for first, named in 159 answers against 148. Meltwater stayed third with 142, followed by Prowly, Semrush and Prezly. Among smaller tools, Onclusive rose from 25 answers to 35 and CoverageBook moved into the top ten, while Qwoted fell from 34 to 23 and PR Newswire from 42 to 32.
A caution: this is one week against one baseline. Assistants don't answer the same question the same way twice, and moves of two or three percentage points for any tool are within that noise. Agentcy's change is large enough to be real; most of the others will need a few more weeks before they mean anything.
The sources the assistants lean on most haven't changed: vendors' own blogs and comparison pages. prezly.com was cited in 67 answers, meltwater.com in 48 and muckrack.com in 47. In this category, a vendor writing honestly about its market is a large part of what the assistants read.
What this suggests for any brand
- Specific beats general. The questions we started winning have one-line answers: a price, a limit, a feature. Write those facts down on a page, in plain words, where an assistant can quote them.
- Cited comes before recommended. If assistants are quoting your guides but recommending your competitors, the content is working and the proof isn't there yet. The next step is other people saying it: reviews, roundups and press.
- Category questions are earned elsewhere. Nobody wins "what is the best PR software?" from their own website. Those answers follow years of third-party coverage.
- Measure the same questions every week. Without a fixed panel and a baseline taken before the change, none of this would be visible.
Next week
We're working on the gaps: a page that answers the AI visibility plus media coverage question directly, clearer agency pricing for AI visibility, and getting Agentcy onto the third-party lists the assistants cite. The next run is on 27 October, and the index will show the result whatever it is. Google AI Mode returned no answers in either run, so it isn't scored yet.
If you want to see what assistants say about your own brand, the free AI visibility checker gives you a first look, and AI visibility tracking runs your own buyer questions every week.