What we tested and how
When buyers ask AI chatbots for software recommendations, the answers they receive determine which vendors enter the consideration set. Metricus conducted an AI visibility report across a random sample of B2B SaaS companies from BuiltIn's top companies list, spanning process mining, project management, cybersecurity, payments, life sciences R&D, and social media management. All were established, well-reviewed products with real market presence.
For each company, we tested over 150 search queries that reflect how real buyers in each category actually ask questions. We logged whether the brand appeared, how it was mentioned, the sentiment, whether the AI used web search (RAG) or training data (parametric), and which competitors surfaced instead.
The visibility scores
| Company | Category | Visibility | Pattern |
|---|---|---|---|
| SaaS 1 | Process Mining | 83% | Appeared in most queries; missed in broader questions |
| SaaS 2 | Lab Software | 75% | Strong when buyer used technical language; weak otherwise |
| SaaS 3 | Project Management | 67% | Dominant in "which tool is best" queries; invisible in industry-specific ones |
| SaaS 4 | Payments | 67% | Appeared for large-company questions; absent for startup/SMB questions |
| SaaS 5 | Social Media Mgmt | 67% | Visible for mid-market; invisible for solo/small-business |
| SaaS 6 | Cybersecurity | 50% | Top pick for one niche question; invisible for most others |
A 33-percentage-point gap between the most visible and least visible company — all well-established, well-reviewed products. The most consistent pattern: every company had at least one type of buyer question where it was completely invisible.
The wording mismatch: the single biggest predictor
What we found across all audits was that one pattern explained more visibility failures than any other: the gap between how a company describes itself and how buyers describe their problem. SaaS 6 (cybersecurity) positions itself as a “managed EDR platform.” Its ideal customers ask about “ransomware protection” and “cybersecurity tools.” When AI generates search queries from a buyer’s question, it uses the buyer’s language. Better-known competitors appear. SaaS 6 does not.
| Company | Buyer Searches For | Company Says | Result |
|---|---|---|---|
| SaaS 6 | “ransomware protection tools” | “managed EDR platform” | Invisible |
| SaaS 3 | “construction project management” | “work management platform” | Invisible |
| SaaS 4 | “payment processor for online store” | “unified commerce platform” | Listed with caveat |
| SaaS 5 | “social media tools for small business” | “comprehensive social management” | Invisible |
RAG vs. parametric: which engine decides
What we found is that roughly 80% of AI responses in our audit were RAG-driven — the AI searched the web in real time, pulled passages, and assembled an answer with citations. Only one company (SaaS 1, process mining) had confirmed parametric presence, where the AI answered from training data with no web search. The practical implication is significant: if your visibility is RAG-driven, content restructuring can show results in weeks. If it depends on parametric knowledge, you are waiting for the next model training cycle — a timeline measured in months.
What AI actually cites
Across 150+ queries, we found that G2 appeared as a cited source in the majority of audits. Gartner dominated enterprise questions. Company-owned blog content ranked well in raw search results, but AI summaries still favored competitors when third-party editorial sources carried more weight. The finding that matters: AI models treat third-party citations as more authoritative than company-owned content during answer synthesis, even when the company’s own content appears in the underlying search results.
What this means for brands
The core finding across every audit: AI does not recommend the best product. It recommends the product it can find using the buyer’s language, from sources it trusts. Market share, funding, even product quality — none of these guarantee visibility. Wording match, third-party citations, and content structure do. Understanding where your brand stands in this new landscape is the first step to changing the outcome.
Methodology: Structured AI visibility reports conducted March 2026 using AI models with web search augmentation. Companies randomly sampled from BuiltIn’s top B2B SaaS list. Over 150 queries generated autonomously. Results anonymized.
Last updated: April 2026