Between 2026-08-27 and 2026-09-25, AIDailyPulse ran 390 AI visibility probes across 30 distinct days on the Perplexity platform. Each probe paired one of 13 B2B software keywords with a prompt designed to surface vendor recommendations. The monitored B2B software vendor was mentioned in 332 of 390 answers, yielding an overall mention rate of 85.1%. The tone across all answers was neutral in 382 cases (97.9%) and positive in only 8 cases (2.1%). The total cost was $2.34, at $0.0060 per probe.
This study is limited to a single AI platform, a single brand, and one tenant. It does not measure Google search visibility, other AI assistants, or competitive mention rates. The data reflects what Perplexity surfaced for these specific keyword prompts during a single month.
Finding 1: The vendor was named frequently, but almost never in a positive frame.
The monitored B2B software vendor appeared in 85.1% of all answers. That sounds strong. But the tone breakdown tells a different story. Only 2.1% of answers carried a positive tone. In 97.9% of cases, the vendor was mentioned neutrally — typically as one option among many, or in passing. High mention rate does not equal positive association. It means the vendor is visible enough to be named, but not compelling enough to be recommended favorably.
Finding 2: Mention rates varied by keyword, but the gap between top and bottom was narrow.
| Keyword | Probes | Mentioned | Rate |
|---|---|---|---|
| smart education solution | 30 | 28 | 93% |
| mini-program development | 30 | 28 | 93% |
| AI visual recognition | 30 | 28 | 93% |
| supply chain digitization | 30 | 27 | 90% |
| AI application development | 30 | 27 | 90% |
| custom ERP CRM OA software | 30 | 26 | 87% |
| smart community management | 30 | 26 | 87% |
| government digitalization platform | 30 | 25 | 83% |
| AI customer service system | 30 | 24 | 80% |
| LLM deployment service | 30 | 24 | 80% |
| digital transformation solutions | 30 | 23 | 77% |
| e-commerce mall system | 30 | 23 | 77% |
| data visualization dashboard | 30 | 23 | 77% |
The top three keywords — smart education solution, mini-program development, and AI visual recognition — all sat at 93%. The bottom three — digital transformation solutions, e-commerce mall system, and data visualization dashboard — sat at 77%. The spread is only 16 percentage points. The vendor is consistently named across the keyword set, but consistently absent from the top of recommendation lists.
Finding 3: The vendor's own domain was not the most-cited domain in AI answers.
The most-cited domains in the Perplexity answers were third-party sources. www.zzcxin.cn led with 705 citations, followed by www.etycx.com at 436 and www.asiainfo.com at 343. Other notable domains included t.cj.sina.com.cn (215), www.linghuicn.com (152), and zzcxin.cn (151). The remaining top domains — www.datamargin.com, www.credamo.com, www.ixcsz.com, and nxinsz.com — rounded out the list with between 77 and 142 citations each.
If the vendor's own domain appears in these results, it is not among the top ten. The AI is citing industry publications, aggregators, and comparison sites far more often than it is directing users to the vendor directly.
Finding 4: Neutral tone dominated every keyword.
With 97.9% of all answers rated neutral in tone, there is no keyword where the vendor broke out of the neutral band. Even the highest-mention keywords — smart education solution, mini-program development, AI visual recognition — did not produce a single positive-tone answer. The vendor is being named, not endorsed.
Finding 5: The cost of measurement was minimal, but the insight is expensive to ignore.
The entire probe set cost $2.34. That is less than a single sponsored click in many B2B channels. Yet the data reveals a structural gap: the vendor is present in AI conversations but absent from AI recommendations. Fixing that gap will cost far more than $2.34.
Most B2B teams treat mention rate as a proxy for AI visibility. This data shows why that assumption is dangerous. A 85.1% mention rate sounds like dominance. In practice, it means the vendor is a footnote in the AI conversation — named, but not trusted enough to be recommended. When Perplexity surfaces a vendor in a neutral tone alongside a dozen other options, the user is unlikely to click through. The vendor is visible, but not actionable.
The domain citation data reinforces the point. The AI is pulling from third-party sources — news sites, review platforms, aggregators — rather than the vendor's own domain. That means the brand is being discussed in other people's contexts, on other people's pages, with other people's framing. The vendor has little control over how it is represented. For a B2B software company, that is a significant loss of narrative control.
For SEO and AEO practitioners, the implication is straightforward. Building mention rate without building recommendation rate is an incomplete strategy. The goal should not be to appear in AI answers. The goal should be to be the answer — or at least the first recommended option. That requires owned content that AI systems can cite directly, positive sentiment signals, and domain authority that competes with the third-party sources currently dominating the results.
1. Audit your owned content for AI-citation readiness. Review your top landing pages, case studies, and product documentation. Are they structured so that an AI system can extract a clear, positive recommendation? If your best content is buried behind login walls or written in vague marketing language, AI systems will cite third-party sources instead.
2. Target the lowest-mention keywords first. Digital transformation solutions, e-commerce mall system, and data visualization dashboard all sat at 77% mention rate. These are the gaps. Invest content and technical SEO effort in these areas before expanding into keywords where you already perform well.
3. Build positive-tone signal sources. The 2.1% positive-tone answers are your proof of concept. Identify what was different in those cases — specific content, a particular domain citation, a different prompt structure — and replicate it. Neutral mentions are easy to generate. Positive recommendations require deliberate signal building.
4. Increase direct domain citations. The data shows third-party domains dominating the citation landscape. Publish original research, benchmark reports, and vendor-agnostic comparisons on your own domain. AI systems cite sources they trust. If your domain is not in their training corpus as a credible reference, they will cite someone else's.
5. Track recommendation rate, not just mention rate. Define a new metric: the percentage of answers where the vendor is not only named but recommended as a top option. Set a target. Measure it monthly. Mention rate is a vanity metric. Recommendation rate is what drives pipeline.
The monitored B2B software vendor was named in 85.1% of Perplexity answers, but in zero percent of cases was it recommended positively. High mention rate without positive recommendation is visibility without influence. The AI conversation about your brand is happening — but it is not happening on your terms. The gap between being named and being recommended is where B2B AI visibility strategies either succeed or fail. This data measures the gap. The next step is closing it.
Run a free checkup to see whether ChatGPT, Perplexity and Gemini name, cite and recommend Mention's industry — and where your own brand stands on AI visibility.
→ Get your free AI visibility checkup
Talk to us