Crawlable, but Not Citable.
We analyzed 304 B2B SaaS companies for AI-search readiness. Almost all are reachable by AI. Almost none are engineered to be recommended. Published June 2026.
let AI crawlers in, access is not the bottleneck
structure their content as answers (FAQ schema)
are invisible when AI is asked to recommend tools in their own category
tools AI names per category, miss the list and you're out of the buyer's consideration set
Your Buyers Already Moved to AI
B2B software research now starts inside an AI assistant more often than on Google. The AI’s shortlist is the new first impression, and if you are not on it, you are not in the conversation.
of B2B software buyers now start research with an AI chatbot, up from 29% a year earlier
of B2B buyers use AI somewhere in the purchase process
chose a different vendor than planned based on AI guidance; 33% bought from a brand they'd never heard of
AI-referred traffic converts far higher than organic (ChatGPT 15.9% vs Google's 1.76%)
They Nailed Access. They Skipped Citability.
Across our sample, the signals that make a site reachable are nearly universal, and even llms.txt adoption is surprisingly high. But the signals that make content quotable, structured answers AI can lift directly, fall off a cliff. Being crawlable is table stakes; almost nobody has done the citability work.
Share of 304 B2B SaaS homepages with each signal. Cyan = access, present everywhere. Red = the answer-structure gap.
A checkbox, not a discipline. Even among the 60% who bothered to publish an llms.txt, just 10% also structure their answers with FAQ schema, barely above the 9% baseline. Adopting the trendy signal is easy; doing the citability work that actually earns a recommendation is not, and almost nobody has.
Invisible in Your Own Category
We asked a leading AI assistant to recommend tools across 22 B2B SaaS categories, in three different buyer phrasings each. It names a generous ~13 tools per category, yet 23% of these prominent companies never appeared for their own category. These are household names in their space. For the thousands of lesser-known tools competing beneath them, the odds are far worse.
A broad shortlist is not a safe one. Thirteen names still leaves most of a category out, and AI rarely volunteers a vendor it has no reason to trust.
of prominent B2B SaaS companies are absent when AI recommends tools in their own category.
Roughly 1 in 4, even among the well-known.
Nobody Is Locking AI Out
Publishers may be blocking AI crawlers, but B2B SaaS is not. Just 4% of our sample disallows a single major AI crawler in robots.txt. The handful that do are the exception, not the trend:
The takeaway is the whole point of this study: the door is open and the basics are done. Visibility in AI is no longer an access problem, it is a citability problem. Getting recommended takes structured answers, third-party authority, and freshness, the work 91% of this sample has not started.
Methodology
- Sample. 304 reachable B2B SaaS companies across 22 categories (martech, devtools, data, HR, fintech, security, support, and more). The sample skews toward prominent, tech-forward companies, so real-world adoption across all B2B SaaS is likely lower than reported here.
- Crawl signals. For each site we fetched the homepage, robots.txt, and llms.txt and recorded AI-crawler rules, structured-data presence and type, and answer-readiness basics. Every crawl figure is a verifiable fact about the live site; llms.txt and crawler-blocking were content-validated to remove false positives.
- AI visibility. For each category we queried a leading AI assistant with three buyer phrasings and recorded which tools it recommended, then matched against our sample. This reflects one capable model at one point in time; engines and answers shift, which is itself the point.
- Honesty. No figure here is modeled or projected. Where a number depends on a single engine, we say so.
Sources
See Where You Land.
Run the free AI Visibility Checker on your own site and see whether AI recommends you, who it names instead, and the citability gaps holding you back, in about 20 seconds.