85% of B2B software buyers think more highly of a vendor when an AI chatbot mentions it in a recommendation.4 A raw mention count treats every mention the same, and says nothing about whether a chatbot, or anyone else, ever read the page yours sits on.
Whether mentions move sales is a fair thing to doubt while the count climbs and organic traffic does not. The published research says mentions move what engines name and what buyers shortlist, and it stops before revenue. The mentions worth chasing sit on a narrower set of page types than the dashboard admits.
Highlights
Mentions track what engines name and what buyers shortlist, and the evidence stops short of revenue. What separates a mention that matters from one that does not is the page it sits on. Listicles, community threads, and review pages get read, and news pages rarely do.
A Brand Mention Is Any Public Reference To Your Name
A brand mention is any place on the web that names your brand, whether or not it links to you. Ahrefs counts both kinds together in its research.1 The unlinked kind can be the bigger one: in one Ahrefs pull for “nike”, 84% of the pages that named the brand carried no link to it.3
The Nike pull covers one brand and one week, on pages filtered to sites with real traffic. It shows that a backlink report can miss most of what is said about a brand.
Mentions also come in different forms, and each form lives somewhere different.
| Form | Where it lives | What the evidence says |
|---|---|---|
| Linked mention | Any page that names and links to you | Backlinks matched AI visibility least of the signals Ahrefs compared1 |
| Unlinked mention | Any page that names you without a link | Counted with linked ones in Ahrefs’ figure, and the text is what language models read1 |
| Press coverage | News sites and trade publications | Built from PR pitches, and rarely cited by engines9 10 11 |
| Listicle or roundup | “Best tools for…” pages | The most cited page type Qvery can name11 |
| Review or directory page | G2, Capterra, and their peers | Level with chatbots as the top shortlist source5 |
| Community thread | Reddit and Q&A sites | Reddit holds most citations to user-generated content13 |
| Video | YouTube titles, transcripts, and descriptions | The strongest single signal Ahrefs tested2 |
| AI answer | ChatGPT, Google AI Mode, AI Overviews | Shifts how buyers rate the vendor named4 |
Seven of those eight are pages other people wrote about you. The eighth is the answer a buyer reads, and the other seven are its reading material.
AI Engines Name The Brands The Web Mentions Most
Ahrefs compared 75,000 brands’ visibility in Google AI Overviews with the signals around them.1 Web mentions matched it best, with a correlation of 0.664. Backlinks matched far less, at 0.218.
A correlation of 1 means two rankings match exactly and 0 means they have nothing to do with each other, so 0.664 is a clear but imperfect match. It is a correlation only, and Ahrefs says correlation is not causation.1
The gap between brands is steep, and it favors whoever is already ahead. The top quarter of brands by web mentions collected more than ten times the AI Overview mentions of the next quarter down, and about a quarter of all 75,000 had none.1
Video follows the same pattern. A brand name in a YouTube title, transcript, or description showed a correlation of 0.737 with visibility across ChatGPT, AI Mode, and AI Overviews, the highest Ahrefs found.2
Search Everywhere Optimization is the workstream where Empact Partners works on this: making a partner visible on every surface a buyer searches, the AI engines and the communities they cite included.
Empact Partners is a consultancy that takes on go-to-market for B2B software companies, and since 2020 we have run that work from inside their own marketing teams. Ahrefs’ numbers tell us where to point it, and they do not tell us what it will return.
.png)
Buyers Pick A Shortlist Before Sales Gets Involved
A mention only pays off if it reaches a buyer before the shortlist closes. 6sense surveyed nearly 4,000 B2B buyers, and 95% of the time the winning vendor was already on the buyer’s first shortlist.6
Chatbots now help build that list. G2 asked 1,076 software buyers in March 2026, and 51% said they start research in an AI chatbot more often than in Google, up from 29% eleven months earlier.4 69% had chosen a different vendor than planned after a chatbot’s guidance.4
Review sites matter as much. In G2’s separate 2026 report, 38% of buyers named review sites as the top source shaping their shortlist and 37% named AI chatbots.5 G2 runs a review site, so weigh the 38% accordingly. The near tie is the point: a mention on either surface reaches the list.
Mentions Change What Buyers See, But No One Has Measured Revenue
Seer Interactive tracked 3,119 informational searches for 42 organizations, and split them by whether an AI Overview appeared and whether the brand was cited in it.7
Brands cited in the answer got more clicks than brands left out of it, and still under half the clicks of a query with no AI Overview. Seer says it cannot prove the mention causes the gap, since brands Google already trusts may simply be the ones it names.
The click is rare either way. In Pew’s browsing panel of 900 US adults, people clicked a link inside an AI summary on just 1% of visits.8 A mention in an answer is read where it appears.
Our own record stops at the same place. Over three months, Empact Partners’ Reddit program for KKday earned 60+ explicit mentions and 3,208,973 views, and the published case study carries no sales or booking figure.14
The chain has a study behind each link and none behind the whole. Ahrefs links mentions to AI visibility, G2 links AI answers to the shortlist, 6sense links the shortlist to the win, and Seer links a mention to a click. Each party measured a different population, and none follows a mention to a closed deal, ours included.
.png)
Sorting mentions by the page they sit on is the one measurement left that is checkable and ours to do.
Press Coverage Is Well Studied And Rarely Cited By Engines
Press coverage is the form of mention with the best data on how it gets made.
That describes how coverage is made, and nothing about whether an engine reads the result.
The reading side comes from Qvery, which tracks which brands ChatGPT and Google AI Mode name and cite. Discount its counts accordingly, and check each at its source under Sources.
Qvery can name the page type of about 12% of its citations, so the shares below count only what it can name.
Press is not worthless. It changes what press is for: a feature in a trade publication reaches that publication’s readers, and it reaches an engine only if the page is one the engine already reads. Empact Partners does not count a press mention as progress toward AI visibility until the page shows up where the engines cite.
Engines Read Lists, Threads, And Review Pages
Ranked by page type, Qvery’s counts put listicles first, then reviews and discussion pages, with news last.11
The engines differ, and software adds Reddit.
A mention on a roundup helps most where ChatGPT is the engine your buyers use.
The list of cited pages is long, though. 46.5% of cited domains appear exactly once.12 The published counts do not support a short list of pages deciding each category. What concentrates is the type of page.
Empact Partners works on that page type through Existing Article Outreach, our method for getting a partner’s product named in the software roundups that already rank and that engines already cite. It is slow work, measured in quarters.
We Count A Mention Only When An Engine Already Reads The Page
The rule Empact Partners runs on a partner’s mention log is short. A new mention counts as progress when it lands on a page, or a kind of page, that an engine already cites in that partner’s category, which we check against the citation list in Qvery.
The rule answers the reliable-source question in both directions. A mention on a page no engine cites may still be read by a buyer, so it stays in the log as reach. A mention on a page an engine cites but no buyer trusts is unlikely to shape a shortlist, and that is our reading and not a measured effect.
Search Everywhere Optimization is the discipline that decides which mentions are worth chasing. Mention building, the counterpart inside our Generative Engine Optimization work, acts on that reading. In practice a partnership runs like this:
Watch One Share Instead Of The Total
The evidence leaves open the things a board most wants settled.
Until someone publishes the missing measurement, watch the share of last month’s logged mentions that sit on pages an engine already cites in your category. If nobody on the team can answer that, the total on the dashboard is not measuring what the board thinks it measures.
If your mention count keeps rising while organic traffic and pipeline stay flat, book a call with us. We will tell you plainly whether this is work we should do together.
Sources
- Ahrefs, “An Analysis of AI Overview Brand Visibility Factors (75K Brands Studied)”, 2025, updated 2026. Spearman correlations for 75,000 brands (domains rated above DR 40 whose top keyword draws at least 800 searches a month) against mentions in Google AI Overview responses, via Ahrefs Brand Radar. Read 21 September 2026.
- Ahrefs, “Top Brand Visibility Factors in ChatGPT, AI Mode, and AI Overviews (75k Brands Studied)”, December 2025, updated 2026. The same 75,000-brand filter, extended to ChatGPT and AI Mode. Read 21 September 2026.
- Ahrefs, “A Simple Guide to Turning (Unlinked) Brand Mentions into Links”, 2018, updated 2020. One author’s worked example: 393 English pages mentioning “nike” in one week, on DR 30+ sites with at least 50 monthly organic visits. Read 21 September 2026.
- G2, “The Answer Economy: How AI Search Is Rewiring B2B Software Buying”, April 2026. Online survey of 1,076 B2B software buyers and decision makers across North America, EMEA, and APAC, March 2026. Read 21 September 2026.
- G2, “2026 Buyer Behavior Report: The Evaluation Maze”, July 2026. Survey of more than 1,000 B2B software buyers and decision makers, with more than 50 B2B sales and marketing leaders interviewed. Read 21 September 2026.
- 6sense, “The B2B Buyer Experience Report for 2025”. Nearly 4,000 B2B buyers across North America, Europe, and Asia-Pacific, 2025. Read 21 September 2026.
- Seer Interactive, “AIO Impact on Google CTR: September 2025 Update”, 2025, updated 2026. 3,119 informational search terms across 42 organizations, June 2024 to September 2025, with third-quarter 2025 averages. Read 21 September 2026.
- Pew Research Center, “Google users are less likely to click on links when an AI summary appears in the results”, July 2025. Browsing data from 900 US adults, March 2025. Read 21 September 2026.
- Cision, “PR Emerges as the Primary Source for Journalists in High-Pressure Newsrooms” (2026 State of the Media Report), May 2026. Survey of 1,899 journalists across North America, EMEA, and APAC, January and February 2026. Read 21 September 2026.
- Muck Rack, “The State of Journalism 2026”. Nearly 1,100 journalists surveyed, 897 reliable responses after cleaning. Read 19 and 21 September 2026.
- Qvery, “Listicles Are the Most Cited Content Type in AI Search”, March 2026. Page types of ChatGPT and Google AI Mode citations, classified from title patterns; the classifier names about 12% of citations. Read 21 September 2026.
- Qvery, “Mentions Are the New Backlinks for SaaS”, June 2026. Millions of ChatGPT and Google AI Mode citations since January 2026. Read 21 September 2026.
- Qvery, “98% of UGC Citations in AI Search Come From Just Four Platforms”, May 2026. ChatGPT and Google AI Mode citations tracked in Qvery’s AI Engine Researcher. Read 21 September 2026.
- Empact Partners, “KKday” case study, empact.partners/partners/kkday. A three-month Reddit program for KKday, as the case study publishes it. Read 21 September 2026.
