Buyers don't shortlist anymore. They ask AI.
When someone asks ChatGPT for "the best tool for X," an answer comes back with three names on it. If yours isn't one of them, you lost a deal you never knew existed. Viz tells you where you stand - and who's beating you.
The funnel moved, and your analytics didn't.
Invisible evaluations
AI assistants compare you to competitors in private conversations. No search console impression, no landing-page visit - the shortlist forms before anyone touches your site.
Your docs feed the models
GPTBot and ClaudeBot crawl your documentation, changelog, and blog. Those pages shape what AI says about your product - and your JavaScript analytics can't see any of it.
Unattributed signups
Trials that start from a ChatGPT or Perplexity recommendation land as "direct" in your dashboards. The channel that's growing fastest is the one you can't measure.
Measure the whole AI funnel, from crawl to signup.
Track your buyer prompts
Monitor the "best X for Y" questions your market asks across Gemini, Claude, GPT, and Perplexity - mention rate, citations, sentiment, and share of voice against the competitors you pick. AI visibility →
See who reads your docs
Server-level analytics shows which AI crawlers hit your content and how often - the leading indicator that models are learning who you are. Analytics →
Attribute AI-driven signups
Connect GA4, and Viz surfaces sessions, engagement, and conversions from visitors AI assistants send you - so the new channel finally shows up in the numbers.
Documentation is your new sales channel - and it sells to machines.
When a buyer asks an AI assistant to compare tools, the assistant doesn't browse your homepage the way a human would. Its answer is assembled from what models learned during training and what retrieval bots fetch at question time - and both lean heavily on documentation, changelogs, comparison pages, and technical blog posts. The content your marketing site treats as an afterthought is often exactly what the machines weigh most.
That changes what "SEO" means for a SaaS company. Classic search optimization competed for position on a page of ten links; AI answers typically name two or three products, and everyone else might as well not exist. Getting into that shortlist is a function of being crawled, being understood, and being cited - three things you can only manage if you can see them happening.
The practical loop looks like this: watch which of your pages AI crawlers actually read (the crawl is the leading indicator), track whether answer engines mention and cite you for your buyers' questions (the outcome), and attribute the trials that AI referrals send you (the payoff). Each step is measurable, and each one tells you where the next content investment should go.
Common questions, straight answers.
How do I know if AI models even know my product exists?
Two signals. First, crawl data: if GPTBot, ClaudeBot, and friends are fetching your docs and marketing pages, your content is entering the pipeline models learn from. Second, answer tracking: run the questions your buyers ask against the major engines and see whether your brand comes back. Viz gives you both, side by side.
Should a SaaS company block AI crawlers?
Usually not the retrieval and search bots - being absent from AI answers is a real commercial cost when buyers ask assistants for recommendations. Training crawlers are a genuine judgment call, but for most SaaS businesses, being known by the models is worth more than the content is. The right answer starts with seeing who crawls you and how often.
Why do AI-referred trials show up as direct traffic?
Many AI surfaces open links without a referrer header, or from apps where no referrer exists. Some, like chatgpt.com, do pass a referrer that GA4 lumps under generic referral traffic. Viz connects to GA4 and pulls the AI-attributable sessions into one view so the channel stops hiding inside "direct."
Can I see which competitor content wins the citations I lose?
Yes. AI visibility tracking records which sources answer engines cite for your tracked prompts - including when the citation goes to a competitor's comparison page or a review site instead of you. That list is effectively your AI-era content roadmap.
Our docs are public anyway - what does measuring the crawl actually change?
It changes what you write next. Public-by-default is a distribution choice; measurement tells you whether it is paying off. If retrieval bots hammer your integration guides but never touch your comparison pages, you have learned where AI thinks your authority ends - and that the content shaping buyer-facing answers is not the content you assumed. That is a concrete basis for reprioritizing your content roadmap, not a hunch.