1 in 3 Volusia County Business Websites Are Invisible to AI Search (2026 Study)

AI & Automation · 7 min read · October 8, 2026

By Heath Daniel, Founder & AI Platform Architect, HD Connex · Published: October 8, 2026

Direct answer: We tested 43 Volusia County business websites the way AI crawlers fetch them: no JavaScript, bot user-agent. Of the 38 we could analyse, 12 (32%) give AI search engines nothing usable — seven return an error to GPTBot while serving browsers normally, five render everything in JavaScript. Six of those seven blocks are accidental.

We tested 43 Volusia County business websites using the same request an AI crawler makes — no JavaScript, bot user-agent. 38 returned enough to analyse. 12 of those 38 (32%) give AI search engines nothing usable. Seven serve an error to GPTBot while serving a browser normally; five render their entire page in JavaScript, which AI crawlers do not execute. Only one of those seven blocks is deliberate.

Why this measurement is different

Most local SEO audits check what Google sees after rendering JavaScript. That misses what now matters: when someone asks ChatGPT, Claude, or Perplexity "who's a good roofer in Ormond Beach," those systems fetch pages with a plain HTTP request. They generally don't run JavaScript, and they give up fast on errors.

So we made that exact request — bot user-agent, no rendering, one shot — and then a second request as a normal browser, to tell "the site is down" apart from "the site refuses AI crawlers specifically."

How many Volusia County sites can AI search read?

There was no middle ground. Zero sites landed between 100 and 299 static words — a site either serves its content in HTML or it serves almost none.

Did these businesses choose to block AI?

Almost never. The seven blocked sites returned 403, 404, 429, or 520 to a single GPTBot request while serving the same page to a browser. Four of the seven returned 429 Too Many Requests or 520 — the signature of a bot filter, not of real traffic. The blocking happens at the edge: a firewall rule, a bot-fighting setting, a rate limiter that treats an unfamiliar user-agent as an attack.

In August we read robots.txt for the 30 sites that answered the bot, and none disallowed GPTBot, ClaudeBot, PerplexityBot, or CCBot. Re-checking all 38 on 8 October, exactly one — a real estate site that also returns 403 to GPTBot — disallows GPTBot and ClaudeBot. That block is a decision. The other six are accidents, and their owners have no way to see them: the site loads perfectly on their phone.

How much structured data do local sites carry?

FAQ schema at 8% is the widest open gap in the county. Question-and-answer markup is among the most directly extractable formats for AI answers, and 35 of 38 businesses have none.

Which fundamentals are still missing?

A third of these businesses have no phone number an AI crawler can read. When an assistant is asked to find a number and call, those businesses are not candidates.

How many sites publish llms.txt?

12 of 38 sites (32%) publish an llms.txt file — far higher than we expected for a county market, which suggests AI-SEO vendors have been working this area.

But one of those 12 publishes llms.txt on a site that renders nothing without JavaScript. The file politely explains the business to an AI crawler that already gave up. Nine of the 12 do it properly: readable HTML, LocalBusiness schema, and the file. llms.txt is a finishing move, not a foundation.

Which industries scored best and worst?

Healthcare and law came out cleanest — both are industries where compliance-minded web vendors are the norm. Real estate and professional services fared worst; both lean heavily on template platforms and agent-branded microsites. The samples per industry are small, so read these as direction, not verdicts.

Methodology

• Sample: 43 Volusia County business websites across seven service industries in Daytona Beach, Daytona Beach Shores, Ormond Beach, Port Orange, New Smyrna Beach, South Daytona, Holly Hill, Edgewater, DeLand, and Deltona, collected from public search results for each industry and city and filtered to independent local operators. National chains and directory pages were excluded.
• Excluded: 5 sites returned 403 to every request from our datacenter IP regardless of user-agent, which we cannot distinguish from a geographic or network block. Reporting them as "blocking AI" would overstate the finding. That left n = 38 (the one restaurant site was among the five, so six industries remain).
• Requests: one GET / with User-Agent GPTBot/1.0 and no JavaScript execution; a paired request with a desktop Chrome user-agent for any non-200; plus /robots.txt and /llms.txt.
• "AI-invisible" means the site returns an error to the bot request while returning 200 to a browser, or yields under 100 words of text in static HTML.
• Measured 26 August 2026. robots.txt was re-checked for all 38 sites on 8 October 2026, because the August run skipped it for sites that blocked the bot request; both results are in the dataset.
• Scoring, out of 100: AI-readable (40), LocalBusiness schema (15), FAQPage (10), any JSON-LD (5), title (5), meta description (5), H1 (5), viewport (5), static phone (5), robots.txt (5). Median across the county: 75/100.

Disclosure: two sites in this sample — floguardfl.com and culverrealtygroup.com — are HD Connex client sites. They scored 100 and 90. They were measured with the identical script and are named here so you can weigh that yourself rather than take our word for the median.

What should a Volusia County business do about it?

1. Test your own site. From a terminal, run curl -sA "GPTBot/1.0" https://yoursite.com/ | wc -w — under 100 means AI search cannot read you. Compare it against a normal browser fetch: if the browser works and the bot doesn't, your host or firewall is the problem, not your content.
2. Ask your host to allow the AI crawlers. On Cloudflare it is a single setting; on most managed WordPress hosts it is a support ticket. Name them explicitly: GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended, CCBot.
3. Get your content into HTML. If your pages are built by JavaScript, your platform almost certainly has a prerender or static-export setting.
4. Add LocalBusiness schema and an FAQ block. 53% of your local competition lacks the first and 92% lacks the second.
5. Put your phone number in the HTML as a tel: link, not inside an image or a script-injected widget.

Sources and further reading

• OpenAI — GPTBot and OpenAI crawlers
• Google Search Central — Google's crawlers, including Google-Extended
• Schema.org — LocalBusiness
• Google Search Central — structured data general guidelines
• The llms.txt proposal

The dataset

Full results — 38 sites, 24 measured fields each, plus the October robots.txt re-check — are published as CSV under CC BY 4.0 so anyone can check the numbers or re-run the study: volusia-web-readiness-2026.csv.

Re-runs are planned quarterly. If your site is in this sample and you have fixed something since 26 August 2026, tell us and we will re-measure it.

Topics: AI Search, GEO, Volusia County, Research, llms.txt, Structured Data

Related: Daytona Beach Web Design · Website Design Packages · Local SEO + GEO Guide (2026) · Get Your Site Checked · Industry solutions (Daytona Beach) · Website design packages

← Back to blog · Industry solutions