Blog · Oct 8, 2026
We scanned 10,369 small business websites — here's how AI-ready they are
When someone asks ChatGPT about your business, the answer comes from what it could read on your website. We wanted to know: how well are real small business sites actually doing?
So we scanned 10,369 of them.
What we did
We ran every site through our scanner: five checks (can AI crawlers reach it, is the page structured clearly, does it have proper titles and descriptions, is the content readable, and does it load fast) plus an FAQ adjustment — sites with helpful Q&A content earn up to 10 bonus points, sites with none lose 10. Each site got a score out of 100. Sites that block AI crawlers can't score above 40.
No cherry-picking — these are real small business websites across the US, drawn from public business listings, on every major platform from WordPress to Wix to Squarespace. The denominators matter, so here they are: we attempted 13,342 sites. 10,369 loaded and were scored — every score figure below is out of those 10,369. The other 2,973 (22.3% of attempts) didn't answer, returned errors, or refused the connection; they were excluded from scoring and are counted separately in the headline numbers further down.
Score distribution
Most sites cluster in the 60-79 range. Only 3.2% score 90 or higher — the "Ready" band, where AI systems can find and read everything on the site. (Readable is not the same as recommended — that also depends on reviews, listings, and the question asked. Our score measures the reading part.)
The headline numbers
- 77.7% of the 13,342 sites we attempted loaded and were scored (10,369 sites)
- 22.3% couldn't be scored at all — dead links, expired domains, server errors, refusals
- 3.4% of attempts (449 sites, counted inside the 22.3%) answered our scanner with a 403 Forbidden error
- 2.3% (303 sites, also inside the 22.3%) required a login (401) — invisible to AI by design
"1 in 68 small business websites blocks every AI crawler we test — often without the owner knowing."
The 403 problem
449 sites returned a 403 Forbidden error. That means the server saw our request and said "no." For AI systems, a 403 is a dead end — they can't read the page or summarize it, so that business simply won't be in the answer.
One honest caveat: a 403 doesn't prove a site deliberately blocks AI. Firewalls and bot protection answer every automated visitor this way. It proves the effect, not the intent — and for an AI system trying to read the page, the effect is what matters.
Here's the thing: most of these owners have no idea. A 403 usually comes from a security plugin, a firewall rule, or a hosting default — not a deliberate choice. The site looks fine in a browser. But to ChatGPT, it's a brick wall.
If your site is one of them, you're invisible in AI search and you don't know it.
What the platforms tell us
WordPress powers the most sites in our sample (3,784), followed by Wix (1,105) and Squarespace (354). Platform choice matters less than configuration — a well-set-up WordPress site beats a neglected Squarespace site every time.
The pattern we see: the platform doesn't determine your score. The details do. A missing meta description here, a blocked crawler there, headings with no hierarchy — small things that add up.
What this means for your business
FAQ is the weak spot
81% of sites score 40 or below on FAQ signals. Most businesses don't have structured Q&A content that AI can easily extract.
A 67 average sounds low — and it is. Every point below 90 is something on a site that AI couldn't fully read. The sites scoring in the 90s aren't doing anything exotic — they just have the basics right.
The good news: most fixes are simple. Unblock your crawlers. Add a meta description. Use real headings. Put your key facts in text, not images. None of this requires a developer.
Check your own site
We built CrawlReady so any business owner can see exactly where they stand. Run a free scan and you'll get your score plus the specific fixes — in plain English, no jargon.
Methodology: 13,342 small business websites attempted in October 2026, drawn from public business listings across the US; 10,369 loaded and were successfully scored using the CrawlReady v2 scorer (final), and every percentage in this article is out of 10,369 unless it says "of attempts." The remaining 2,973 (22.3% of attempts) didn't answer, returned errors, or refused the connection and were excluded from scoring. Five weighted checks (crawler access 20%, structure 25%, metadata 25%, content 15%, performance 5%) plus FAQ adjustment (-10 for FAQ score ≤40, +10 for FAQ score >70). Sites whose crawler access scores 0 — every AI crawler we test is turned away — are capped at 40. Score bands: Ready 90+, Workable 70-89, Needs Work below 70. All scores rescored with v2 on 2026-10-11. Scores reflect what AI systems can read, not traditional SEO rankings, and readability alone never guarantees a citation or recommendation.