By Steve Merrill, Founder of WRKNG Digital | November 11, 2026
What did WRKNG find when it audited 50 B2B SaaS sites for AI visibility?
44 of the 50 B2B SaaS sites we audited had zero FAQ schema anywhere on their domain. 38 had pricing pages with less extractable text than a single product description on Amazon. Only 6 showed up when we asked ChatGPT and Perplexity for "best [category] software" in their own category. That gap is not random. It tracks almost exactly with which sites use structured data and which ones don't.
We built our AI-visibility audit on Shopify stores first. Clean product schema, FAQ markup, feed hygiene, that kind of thing. This month we ran the same checklist against 50 B2B SaaS marketing sites: homepages, pricing pages, and case study pages. This is WRKNG's first pass into B2B. We don't have years of track record here yet. What we have is the same methodology that's worked on ecommerce, pointed at a new kind of site, with the receipts to show for it.
How we picked the 50 sites
We pulled a mix of B2B SaaS companies across sales tech, marketing tech, HR tech, and dev tools. Mid-market tools, not just the household names. Each site got the same three checks: pricing page content depth, case study structured data, and FAQ schema presence. Then we ran six to eight category queries per vertical through ChatGPT and Perplexity, the kind a buyer actually types: "best project management software for agencies," "top CRM for small sales teams."
Same approach I'd use pulling apart a Shopify product feed. Look at what's actually on the page in machine-readable form, not what the marketing team thinks is there.
Why don't most B2B pricing pages show up in AI answers?
Because there's nothing for an AI model to extract. 38 of 50 pricing pages used vague tier names ("Starter," "Growth," "Enterprise") with no visible numbers, feature lists hidden behind JavaScript toggles, or a single line: "Contact us for pricing." An AI assistant can't cite a number that isn't on the page.
Not great. That's the plain version of it.
Compare that to the 12 sites with real pricing tables: dollar figures, per-seat costs, a bullet list of what's included at each tier, rendered in plain HTML rather than buried in a script. Those 12 showed up in AI answers at more than double the rate of the other 38. The AI isn't guessing at value. It's reading a table and repeating what's in it.
What's wrong with B2B case studies that AI can't use?
Almost none of them have structured data, and most bury the actual result in a paragraph rather than a headline number. We found exactly 3 of 50 sites using Article or Review schema on their case study pages. The other 47 had case studies written as long-form narrative with the outcome mentioned once, halfway down, in a sentence like "the team saw significant improvement."
Significant improvement is not a data point. An AI model looking for proof can't quote a vague claim. It needs a number: "cut onboarding time by 34%," "reduced support tickets by 1,200 a month." Only 9 of the 50 sites led a case study with a number in the headline or first sentence. Those 9 were cited by name in our Perplexity test queries. The other 41, almost never.
Do "trusted by" logo walls actually help AI visibility?
No, not on their own. 46 of 50 homepages had a "trusted by" section showing customer logos with zero supporting text: no customer count, no named use case, no link to proof. A logo is a picture. AI assistants read text. A wall of logos with no copy around it gives the model nothing to cite.
The 4 sites that paired logos with a specific claim ("Used by 3,200+ sales teams" or "Trusted by teams at companies processing $50M+ in deals") gave the AI something concrete to repeat back. Small fix. Most teams just haven't done it.
What pattern separates the 6 sites that show up in ChatGPT and Perplexity?
All 6 shared three things: FAQ schema on at least one page, a pricing page with real numbers in plain HTML, and case studies with a quantified result in the first two sentences. None of the 6 had every box checked across their whole site. They just had enough of it, in the right places, that an AI model could find a clean answer and quote it.
Here's the thing. None of these 6 are the biggest brand names in their category. A couple are mid-market tools most people outside the industry haven't heard of. Brand size didn't predict visibility. Page structure did.
This matches what Google's own structured data documentation has said for years: machine-readable markup helps search systems understand and surface content. AI answer engines are reading the same signals, just routing them into a chat response rather than a blue link. Gartner has flagged this shift directly, predicting traditional search volume will keep dropping as AI-generated answers take over more buying research.
Why does FAQ schema matter this much for B2B sites?
FAQ schema gives an AI model a pre-packaged question-and-answer pair it can lift straight into its response. Without it, the model has to guess at your answer by parsing paragraphs. With it, you're handing over the exact sentence you want quoted.
44 of 50 sites we checked had no FAQ schema at all, per schema.org's FAQPage spec. Not buried wrong, not broken. Missing. That's the single easiest fix in this whole audit, and almost nobody has done it.
If you do three things this week
I've run this exact audit on 50+ ecommerce sites before turning it on B2B, and the fixes are almost always smaller than people expect. Here's where to start.
One. Put real numbers on your pricing page. Dollar amounts, per-seat costs, a plain HTML feature table. If your pricing is genuinely custom, say so, but still show a range or a starting price. "Contact us" with nothing else is invisible to an AI model.
Two. Rewrite your top 3 case studies to lead with a number. First sentence, not paragraph four. "Cut deployment time by 40% in six weeks" beats "achieved significant efficiency gains" every time, with every reader, human or AI.
Three. Add FAQ schema to your homepage and pricing page. Five to eight real questions your sales team actually gets asked. Mark them up with FAQPage schema so AI assistants can lift the exact answer.
None of this takes a redesign. It takes a week and someone willing to replace vague language with specific numbers.
FAQ
What is an AI-visibility audit for a B2B SaaS site?
It's a review of whether ChatGPT, Perplexity, and similar tools can read, extract, and cite your pricing, case studies, and FAQ content. We check structured data, page content depth, and run real buyer queries to see if your brand gets named.
Why did WRKNG expand from Shopify audits into B2B SaaS?
We built our AI-visibility methodology on ecommerce stores first. The underlying question is the same for any website: can an AI model extract a clean, specific answer from your page? That question applies just as much to a pricing page as it does to a product page, so we pointed the same audit at B2B.
How long does it take to fix FAQ schema and pricing page gaps?
For most of the 50 sites we looked at, the technical part is a day or two of work. Writing honest, specific answers and real numbers for the pricing table usually takes longer than the markup itself.
Does company size affect whether AI assistants cite a B2B site?
Not in this audit. The 6 sites that showed up in our ChatGPT and Perplexity tests were a mix of sizes. What they shared was page structure: FAQ schema, visible pricing numbers, and quantified case study results. Smaller tools with clean data beat bigger brands with vague pages.
What's the single fastest fix for a B2B marketing site?
Add FAQ schema. 44 of the 50 sites we audited had none. It's the cheapest fix in this whole list and it gives AI assistants a direct, quotable answer instead of a paragraph to guess from.
If you want to know where your own site stands, that's what we do now across ecommerce and B2B. See how at wrkngdigital.com/agentic-commerce-landing-page.

