AI brand mentions reached 86.4% across 390 Perplexity probes

LLM

AI brand mentions reached 86.4% across 390 Perplexity probes

What we measured

AIDailyPulse analysed 390 AI probes run from 2026-09-14 to 2026-10-13, covering 30 distinct days. Every probe was submitted to Perplexity using a keyword set of 13 B2B software and digital-services queries. The monitored B2B software vendor appeared in 337 of 390 answers, producing an overall mention rate of 86.4%. Each probe cost $0.0060, for a total measured cost of $2.34.

This is a focused measurement, not a universal benchmark. It covers one AI platform, one monitored B2B software vendor, one tenant, and one keyword set. The results show how often the vendor appeared in this specific Perplexity sample; they do not establish performance across other AI engines, markets, prompt types, or brands. The probe data also records whether the brand appeared and the answer tone, but does not by itself measure position, recommendation strength, traffic, leads, or conversions.

What the data shows

#### 1. The monitored B2B software vendor appeared in most answers

The central result is straightforward: the vendor was mentioned in 337 of 390 answers, or 86.4%. That means the brand appeared in roughly eight or nine out of every ten sampled answers across the 30-day period.

The result is high, but it is not universal. The vendor was absent from 53 answers. That gap matters because a strong aggregate rate can still hide weaker performance for particular services or buyer questions.

The tone data provides additional context:

Most answers were neutral rather than positive. In this study, a brand mention therefore usually meant inclusion in an informational answer, not an explicitly favourable endorsement. The 26 positive answers show that positive framing occurred, but it was much less common than neutral presentation.

#### 2. Visibility was strongest for smart education and AI visual recognition

Two keywords produced the highest mention rate: smart education solution and AI visual recognition. The monitored B2B software vendor appeared in 29 of 30 probes for each keyword, giving both a 97% mention rate.

Smart community management followed at 28 of 30 probes, or 93%. These results indicate that the vendor’s visibility was not evenly distributed across the keyword set. It was most consistently associated with a small group of specific solution categories.

#### 3. LLM deployment service was the clearest visibility gap

The lowest rate in the sample was for LLM deployment service. The vendor appeared in 22 of 30 answers, or 73%.

That is 24 percentage points below the 97% rate recorded for the two leading keywords. The difference suggests that visibility may depend strongly on the wording and category context used by the buyer. A vendor can be highly visible for established service areas while appearing less often for a newer or more technically specific category.

The next-lowest rates were 80% for digital transformation solutions, government digitalization platform, and data visualization dashboard. Each produced 24 mentions from 30 probes. These categories were visible, but less consistently than AI visual recognition, smart education solution, or smart community management.

#### 4. The keyword set produced a broad but uneven visibility profile

The results divide into several practical bands:

• 97%: smart education solution; AI visual recognition
• 93%: smart community management
• 90%: AI customer service system; AI application development
• 87%: mini-program development; custom ERP CRM OA software; supply chain digitization
• 83%: e-commerce mall system
• 80%: digital transformation solutions; government digitalization platform; data visualization dashboard
• 73%: LLM deployment service

Six of the 13 keywords produced mention rates of 87% or higher. Four produced rates between 80% and 83%, while one fell to 73%. This spread is useful because it shows why a single overall visibility score is not enough for optimisation. The aggregate 86.4% result combines categories where the vendor appeared almost every time with categories where it was absent in more than one-quarter of probes.

#### 5. The citation environment was concentrated around a small group of domains

AI answers cited a range of domains, but the largest citation totals were concentrated among the leading sources:

The vendor's own domain, where it appeared in the data, should be distinguished from these most-cited domains: the table identifies domains cited in the sampled AI answers, not a complete account of every source relationship or citation position.

The citation totals also show that domain variants were counted separately in the supplied data. For example, www.zzcxin.cn and zzcxin.cn appear as separate entries, as do qiye.qizhidao.com and www.qizhidao.com. These figures should therefore be read exactly as recorded rather than combined into assumed domain-level totals.

Why it matters for your brand

For B2B marketers, an 86.4% mention rate is a useful visibility baseline, but it should not be mistaken for market-wide awareness or buyer preference. AI engines may mention a vendor without recommending it, explaining its services in detail, or directing a user to the vendor's own domain. The tone distribution reinforces that distinction: 93.3% of all answers were neutral, while only 6.7% were positive. A measurement programme should therefore track mention rate alongside tone, the specific claims made, cited sources, and whether the vendor appears in the part of the answer that buyers are most likely to act on.

For SEO and AEO teams, the keyword differences point to a practical content priority. The monitored B2B software vendor was highly visible for smart education solution and AI visual recognition, but appeared in only 22 of 30 LLM deployment service answers. That gap can guide an investigation into terminology, service pages, technical documentation, third-party references, and evidence that supports the vendor's relevance to the weaker category. The data does not prove that any one content change will increase visibility, but it does identify where future tests should focus.

How to act on it

1. Build a keyword-level baseline.

Record mention rate separately for every target query instead of relying only on the 86.4% aggregate. Keep the 30-probe count for each keyword visible so that changes can be interpreted against the same measurement structure.

2. Prioritise the weakest category first.

Start with LLM deployment service, which recorded a 73% mention rate. Review whether the vendor's relevant pages use the same language as the query, clearly describe deployment capabilities, and provide specific technical information that an AI engine can associate with the category.

3. Protect the strongest associations.

Continue monitoring smart education solution and AI visual recognition, both at 97%. High visibility should be treated as a baseline to preserve, not as a reason to stop measuring.

4. Separate visibility from sentiment.

Track whether answers are neutral or positive. The current sample contains 364 neutral answers and 26 positive answers, so a rise in mentions alone would not show whether AI-generated descriptions are becoming more favourable.

5. Audit cited sources and domain variants.

Review the domains appearing most often in AI answers, including www.zzcxin.cn, qiye.qizhidao.com, and t.cj.sina.com.cn. Check how the monitored B2B software vendor is represented in relevant sources, while keeping domain variants separate unless the measurement rules explicitly combine them.

Bottom line

Across 390 Perplexity probes over 30 days, the monitored B2B software vendor was named in 86.4% of answers, with visibility ranging from 97% for the strongest keywords to 73% for LLM deployment service. The result is a clear baseline: broad AI presence, mostly neutral framing, and specific category gaps that can be measured and addressed over time.