Monolith3
AI Visibility Check / Data report

How visible are Orange County businesses to AI search?

Quick answer: We checked the home pages of 98 of 100 randomly drawn Orange County businesses on . 33 of 98 (34%) kept at least one AI search crawler from reading the page. That's an upper bound: for 8 of 98 (8%) the block is on the page itself and holds no matter who asks.

These are totals only. No business is named here, and nobody's individual result is published.

Updated

Findings

What the crawlers got.

Sample: 100 businesses drawn, 98 scanned, 2 home pages that didn't load for a browser left out. Percentages are of the 98.

  • Most pages were readable

    • 62 of 98 (63%): every AI citation and search crawler we tried got the same page a browser gets.
    • 3 of 98 (3%): they got it, with a warning such as thin text, a slow response or a redirect.
    • 33 of 98 (34%): at least one of them was blocked or got no readable page.
  • Why that 34% is a ceiling

    We imitated each crawler by its user-agent from one ordinary connection, not from the vendors' own networks. 23 of the blocked sites were stopped at the CDN or firewall, and a CDN that checks crawler IP addresses can challenge a fake crawler it would let through for real. So some of those 23 may be readable to the real thing.

    The floor is 8 of 98 (8%). Those sites block crawlers in robots.txt, mark the page no-index, or serve a JavaScript shell with no readable text. That holds whoever asks.

    Before publishing, we re-ran five of the blocked sites by hand, picking different causes. All five came back blocked for the same reason. One let Google's crawler through but turned away the assistants people use to fetch a page live, which looks deliberate.

  • What blocked them

    A site can have more than one cause, so these add up past 33.

    Blocked sites by cause, of 98
    CauseSites
    CDN or firewall (challenge, 403, 429)23
    No-index tag on the page5
    Other errors served to crawlers4
    robots.txt disallow3
    JavaScript shell, no readable text1
  • Which crawlers got blocked

    Sites that blocked each crawler, of 98
    CrawlerWhat it feedsBlocked
    ChatGPT-UserChatGPT fetching a page live30
    Claude-UserClaude fetching a page live29
    PerplexityBotPerplexity's index28
    GooglebotGoogle Search, AI Overviews28
    ApplebotSiri and Spotlight28
    OAI-SearchBotChatGPT Search index27
    Perplexity-UserPerplexity fetching a page live27
    bingbotBing, Copilot27
    DuckAssistBotDuckDuckGo's answers26

    Training crawlers like GPTBot and CCBot were checked too. Blocking those is a policy choice, so it never counts as a failure here.

  • Search basics were thin

    78 of the 98 home pages let our audit tool in. Each was run through the same launch checklist we use on our own sites. None of the 78 passed all of it.

    Home pages that passed each check, of 78
    CheckPassed
    Structured data (JSON-LD) that validates5 (6%)
    Open Graph image for link previews7 (9%)
    Meta description, 50 to 160 characters10 (13%)
    Canonical URL21 (27%)
    Exactly one H1 heading23 (29%)
    Title tag, 15 to 65 characters38 (49%)
    HTTPS on every resource46 (59%)
    robots.txt present49 (63%)

    Every check, with pass and fail counts, is in the totals file.

Method

How we drew the sample and ran the checks.

  • The sample

    • Source: a purchased list of Southern California businesses, built for cold email. We kept the Orange County rows: the county's 34 cities plus unincorporated places.
    • Each business needed its own website domain, one row per domain. That left 27,572 businesses in 25 industries.
    • We drew 100, stratified by industry with proportional allocation, random seed 20261103, on 3 November 2026. We checked each home page.
  • Where the sample is biased

    Businesses with no website aren't on the list at all, and a list built for email leans to businesses that are easy to find online. Both push the numbers toward the better end. The real share of Orange County businesses that AI search can't read is probably higher.

  • AI access

    Each home page was fetched once as a browser and once as each AI citation, search, live-fetch and training crawler. robots.txt was read for each crawler's token. A page passes when every citation and search crawler gets the same readable page the browser gets. It fails on a robots.txt block, a CDN or firewall block, a no-index tag, or a JavaScript shell with no readable text.

  • Search readiness

    Each home page that let the audit in was run through our launch checklist: title, meta description, canonical, structured data, robots, sitemap, headings, alt text, accessibility basics and more. 20 pages refused or failed the audit request and are left out of that part.

  • Same tools as the paid check

    These are the automated checks behind our $99 AI Visibility Check, run unchanged. The paid check adds a person reading the live page. That part wasn't done here.

  • How to cite this

    Monolith3, "How visible are Orange County businesses to AI search?", 3 November 2026. https://monolith3.studio/ai-visibility-check/orange-county-2026/

    Please say the 34% is an upper bound when you quote it. The totals, method and sample notes are in totals.json. Questions about the data: info@monolith3.studio.

    The data is free to reuse under CC BY 4.0. Credit Monolith3 and link this page.

FAQ

Questions about the data.

  • Can I see which businesses were checked?

    No. We publish totals only. A business's own result goes to that business privately, if at all.

  • Why is 34% an upper bound?

    We imitated crawlers from one ordinary connection. Some CDNs check where a crawler really comes from and challenge the imitation, while the real crawler would get through. The 8% floor counts only blocks set on the page itself.

  • Is 98 businesses enough?

    It's enough to show the size of the problem, not to rank industries or cities. At this size, a share near a third carries a margin of roughly nine points either way.

  • Does blocking AI training crawlers count against a site?

    No. Keeping your pages out of model training is a fair policy choice. We only count crawlers that fetch a page to quote it in an answer or to index it for search.

  • Will you run it again?

    If we do, it will use the same method so the numbers line up, and each run gets its own dated page. This one stays as published.

  • How do I check my own site?

    The $99 AI Visibility Check runs these same checks on your site, adds a person reading the live page, and ranks the fixes.

Your site

Find out which side of the line you're on.

The $99 AI Visibility Check runs these checks on your site and ranks what to fix.

Tell us what's broken