Mention Rate Is Not Visibility: 85.9% of AI Answers Named the Vendor

AI visibilityChatGPTLLMRAG

Mention Rate Is Not Visibility: 85.9% of AI Answers Named the Vendor

What we measured

From August 24 to September 22, 2026, we ran 390 AI visibility probes across 30 distinct days using Perplexity as the sole AI platform. Each probe targeted one of 13 keyword phrases relevant to the monitored B2B software vendor, with 30 probes per keyword. The total cost for the full study was $2.34, or $0.0060 per probe.

This study has three limitations worth stating upfront. First, it covers only one AI platform—Perplexity—so the results do not represent Google AI Overviews, ChatGPT, Claude, or other generative systems. Second, it measures a single brand across a fixed keyword set, meaning the findings describe one vendor's visibility in one context, not an industry-wide baseline. Third, the data captures mention presence and tone but does not measure recommendation strength, positioning relative to competitors, or the downstream effect on click-through behavior.

What the data shows

1. The vendor achieved a high overall mention rate, but that rate masks uneven performance across keywords. The monitored B2B software vendor was mentioned in 335 of 390 answers, yielding an overall mention rate of 85.9%. However, the per-keyword breakdown reveals a spread from 97% down to 77%—a 20-point gap between the strongest and weakest keyword performance.

2. The top three keywords account for the bulk of the vendor's visibility, while the bottom three sit 20 points below the average. Keywords with mention rates at or above 90%—mini-program development, smart education solution, and AI visual recognition—collectively produced 85 mentions across 90 probes. The three lowest-performing keywords—digital transformation solutions, e-commerce mall system, and data visualization dashboard—each reached only 77%, meaning the AI system omitted the vendor in roughly one out of every four answers for those terms.

3. The tone of mentions is overwhelmingly neutral. Across all 390 probes, 383 answers (98.2%) carried a neutral tone, while only 7 answers (1.8%) were classified as positive. This distribution suggests that even when the vendor is named, the AI system does not consistently associate it with favorable language. A brand that is mentioned but never recommended occupies a different position in the answer than a brand that is both named and endorsed.

4. The vendor's own domain appears far behind several competing domains in AI citations. The monitored B2B software vendor's own domain received fewer citations than at least six other domains in the same answers. The leading domain, www.zzcxin.cn, accumulated 630 citations across the study, followed by www.etycx.com at 439 and www.asiainfo.com at 350. Domains such as t.cj.sina.com.cn (179), www.datamargin.com (170), and www.linghuicn.com (163) all outperformed the vendor's own domain in raw citation count. This gap indicates that when the AI system references a domain in connection with these keywords, it is more likely to surface a competitor or third-party source than the vendor's site.

5. The study cost was minimal, but the insight it provides is structural rather than marginal. At $0.0060 per probe and $2.34 total, the financial barrier to running this kind of measurement is low. The value lies in revealing a pattern that vanity metrics obscure: a brand can be named frequently and still fail to occupy the recommended position in AI-generated answers.

Why it matters for your brand

For B2B brands, the distinction between being mentioned and being recommended is the difference between awareness and consideration. An AI system that names your brand in a neutral tone is doing the minimum—acknowledging that your brand exists in the topic space. An AI system that recommends your brand is moving your brand into the decision path. The data from this study shows the monitored B2B software vendor sits firmly in the first category. An 85.9% mention rate sounds strong, but the neutral tone distribution and the domain citation gap reveal that the brand is present without being preferred.

This pattern is not unique to this vendor. It is the default state for most B2B brands in AI visibility. Search engine optimization has historically rewarded ranking position. AI engine optimization requires something harder: earning the recommendation slot inside the generated answer. A brand that ranks first on a SERP but is named only in passing by an AI system has not captured the same value. The real metric is not whether the brand appears, but whether the brand appears as the answer to the question, not merely as a footnote within it.

How to act on it

1. Audit your keyword coverage at the per-keyword level. An aggregate mention rate hides weak spots. Identify which keywords fall below your overall average and treat those as priority gaps. The monitored B2B software vendor, for example, dropped to 77% on three keywords while reaching 97% on another. The fix starts with understanding where the brand is being omitted entirely.

2. Build domain authority signals that compete with the domains already winning citations. www.zzcxin.cn, www.etycx.com, and www.asiainfo.com are the domains the AI system surfaces most often. These are not random selections—they are the domains with the strongest signal in the training and retrieval layers of the platform. If your own domain is not among the top citations, the path to improvement is the same one that works for traditional SEO: consistent content publication, authoritative backlinks, and structured data that makes your domain the most citable source for your topic area.

3. Shift your content strategy from mention-worthy to recommendation-worthy. Neutral mentions come from brands that exist in the knowledge graph. Recommendations come from brands that are repeatedly validated by sources the AI system trusts. Publish original research, case studies, and technical documentation that other authoritative domains reference. When third-party sources cite your brand, the AI system is more likely to move from naming to recommending.

4. Monitor tone alongside mention rate on every measurement cycle. A brand that improves its mention rate but sees its positive tone percentage stay flat is not gaining real visibility. Set a target for positive tone percentage, not just mention percentage. Track both metrics in parallel and treat a rising mention rate with falling tone as a warning signal.

5. Run this measurement regularly and on multiple platforms. A single 30-day probe on one AI system gives a snapshot, not a strategy. Repeat the study quarterly, expand to additional AI platforms, and compare results over time. The goal is not a single high mention rate—it is a trend line that moves toward recommendation.

Bottom line

An 85.9% mention rate is not a visibility win. It is a visibility floor. The monitored B2B software vendor is being named, but it is not being recommended, and its own domain is being cited less frequently than several competitors. For B2B brands, the gap between mention and recommendation is where real AI visibility is won or lost. Measuring that gap is the first step. Closing it requires the same discipline that has always driven organic growth: authoritative content, third-party validation, and consistent optimization of the signals AI systems use to decide what to recommend.