We audited 250 B2B SaaS websites for AI readiness. 89% had the traditional SEO groundwork done: sitemap in place, meta descriptions written, the basics handled.
But only 8% had FAQ schema.
That second number is the one that decides whether an AI system can pull a fact off your page and attribute it to you. Which means most B2B websites right now are perfectly findable and almost impossible to cite.
You can see what that looks like in practice. A buyer in your category opens ChatGPT and asks which vendors they should look at. You rank on page one of Google for the same question. Your content is good, your positioning is clear, and you don't get named.
Nothing is broken. Search engines reward pages that rank. Answer engines reward facts they can lift cleanly and credit to someone. Those are two different jobs, and almost everyone has only prepared for the first one.
Here's the rest of what we found, and what to do about it.
What we audited, and why
Of the 250 sites, 150 were built on Webflow and 100 were built on other platforms, including WordPress, Framer, HubSpot, Astro, and custom Next.js builds.
For each one we checked the signals AI systems actually use when they read a site:
- Structured data, and which types were present
- Whether the meta description exists and fits within the length search engines display
- Whether there's a working sitemap
- Whether there's a genuine llms.txt file rather than a placeholder
- Whether robots.txt blocks any major AI crawler
We ran this because we kept seeing the same gap in client audits and couldn't tell whether it was our clients or the whole market. It turned out to be the whole market.
What we found
The pattern was consistent, and the two headline numbers sit right next to each other.
89% have a sitemap and a meta description in place. The traditional SEO groundwork is largely done. Most teams worked through something like a Webflow SEO checklist years ago and have kept it up since.
Only 8% use FAQ schema. This is the markup that tells an AI system "here is a question, and here is the answer to it." It's the closest thing there is to handing a model a quotable fact with your name attached.
A few more findings worth sitting with:
- Of the 172 sites already using structured data, just 12% include FAQ markup. So this isn't a general knowledge gap about schema. Teams know what structured data is. They're skipping this specific piece.
- 31% have no structured data at all. No machine-readable description of what the company is or what it sells.
- 54% have a working llms.txt file. Common enough to be table stakes now, uncommon enough that doing it properly still counts for something.
- Only 7% block any major AI crawler, and not a single site in the study blocked Perplexity's. So access isn't the bottleneck. AI systems can reach almost every one of these sites. They just can't do much with what they find.
- 6 sites out of 250 passed every check. That's 2.4%.
Read together, those numbers say one thing. The old checklist is finished. The new one has barely started.
The finding we didn't expect
We're a Webflow agency, so the honest thing to say is that we expected Webflow sites to score better. They didn't, really.
Webflow sites averaged 3.65 out of 5. Sites on other platforms averaged 3.50. That difference is too small to be meaningful, and it held up when we grouped the sites by company size and funding stage.
Where Webflow did lead was in things the platform handles for you. Every Webflow site in the study had a meta description, and 95% had a sitemap, because Webflow generates one by default. Where it fell behind was Organization schema, which takes a deliberate decision to implement: 51% on Webflow versus 64% elsewhere.
That's the useful takeaway. Platforms give you defaults. Everything past the defaults is a choice your team makes or doesn't. So if you're weighing up a replatform to fix AI visibility, the data says it won't fix it.
What this means if you run B2B marketing
You're probably invisible, not badly ranked
These are different problems with different fixes, and they're easy to confuse.
If your traffic is steady but you're never named in AI answers, more content won't help and neither will more backlinks. The issue is that your pages don't offer anything an answer engine can lift and attribute. That's a structural problem, which means it gets solved in your technical SEO layer rather than your content calendar.
Your competitors haven't figured it out either
92% of the sites we checked are missing the same thing. In a mature channel like SEO, closing a gap like that takes years and a real budget. Here it's markup.
That window won't stay open. Right now doing the basic work puts you ahead of nearly everyone in your category, which is a strange and temporary situation.
The bar keeps moving
More buyers start using an AI tool every month, and the sites that get cited today are training the habit of citing them tomorrow. A site that's readable now compounds. A site that isn't, stays quiet while the traffic it used to get gets summarized somewhere else.
So what is AI readiness, exactly?
AI readiness is how easily an AI system can find your website, understand it, and quote it correctly.
It comes down to four things:
| What it is |
Why it matters |
| Structured data |
Tells AI what your pages contain and what your company does, in a format it can parse without guessing |
| Accurate metadata |
Gives it a clean, correctly sized summary of each page |
| Sitemap and llms.txt |
Tells it what exists on your site and what's worth reading |
| Crawler access |
Makes sure your robots.txt isn't quietly blocking the systems you want to be cited by |
Readiness is a state. It's what an audit measures on a given day.
The ongoing work of getting there and staying there is what's increasingly called AI enablement, which covers the structure, the context AI tools need about your brand, and the rules about what ships without a human looking first. Readiness is the score. Enablement is the practice.
How do you improve your site's AI readiness?
Four fixes, ordered by leverage. These are the ones that move the score most for the least work.
1. FAQ schema on your highest-intent pages
Start with pricing, comparison and alternatives pages, and your top product pages. These are the pages buyers ask AI tools about, and they're the ones you most want quoted accurately.
92% of the sites we checked skip this. It's markup only, so nothing about the page changes for a human visitor.
2. Organization schema sitewide
This is the block that tells AI systems what your company is, what it does, and how to refer to it. It belongs in a template, not on individual pages, so it's usually a one-time job that covers the whole site.
Just over half the Webflow sites we checked had it.
3. Meta descriptions that actually fit
28% of the Webflow sites in our study had meta descriptions longer than 160 characters, which means they get cut off. The median was 151, sitting right at the edge.
This is the least glamorous fix on the list and one of the fastest.
4. A real llms.txt
Nearly half the sites we checked don't have one, and some of those that do are serving a placeholder rather than a curated file.
If you want the step-by-step version of all four, our Webflow AEO Starter Kit has the JSON-LD templates, an llms.txt starter, and a checklist you can work through line by line.
One more thing worth knowing before you start: most of this lives at the template level. On a site where 30 pages share a template, fixing the template fixes 30 pages. Do the template work first and the page-by-page cleanup after.
Why isn't a one-time audit enough?
Because a website changes and its readiness score changes with it.
We ran our checks on a single day in August 2026. Some of those sites will have shipped new pages the following week, edited a dozen others, and updated a template in between. An audit is accurate the day it runs and slowly stops being true after that.
Websites aren't static, so readiness isn't either. New pages launch without schema. Someone rewrites a meta description and pushes it past the limit. A CMS template gets duplicated and the markup doesn't come with it.
Doing this work once is an audit. Keeping it true is enablement, and that's the part that actually protects your visibility.
Where to start
Find out where you currently stand.
Our free AI Readiness Audit checks the same signals this study measured, including structured data coverage, metadata, llms.txt, crawler access, and site health. You'll get a score, the specific issues holding it down, and a note on which fixes cover the most pages at once.
It takes a minute and doesn't require a call.
.png)
.avif)