We measured the homepages of 52 business websites across eight English-speaking Caribbean territories on 31 July 2026. Transport hygiene is close to universal. Machine readability — the thing search engines and AI assistants depend on — is not.
21 of the 52 measured homepages carried no JSON-LD. Structured data is how a search engine or an AI assistant reads a business as an entity rather than as prose, so those sites are relying on the model guessing correctly.
Only 5 homepages used LocalBusiness (or a subtype such as Resort or ProfessionalService) — the one schema type that ties a business to an address, opening hours and a service area.
Not one of the 52 homepages carried FAQPage markup. Answer-shaped content is what AI assistants quote, and nobody in this sample is providing it in machine-readable form.
Median time to first byte across the sample was 0.70s, but a third of sites exceeded 1.0s — before a single image had begun to download.
Measured across HTML plus the first 40 assets. A quarter of sites exceeded 3 MB on that partial count, which is a real cost on Caribbean mobile data.
Only 38% served a sitemap at the conventional path, and 17% served no robots.txt either — the two files crawlers look for first.
All figures measured 31 July 2026 across 52 homepages.
Each homepage was requested once with a mobile user agent over a fixed broadband connection, following redirects, with a 20-second timeout. We recorded the response status, TLS validity, time to first byte, HTML size and the bytes of up to 40 referenced images, scripts and stylesheets, then parsed the HTML for viewport, title, meta description and JSON-LD structured data. robots.txt and sitemap.xml were requested separately.
55 sites were sampled and 52 returned usable measurements. Three sites could not be measured: two failed DNS resolution and one presented an invalid TLS certificate at the time of the scan.
| Territory | Sites measured | Has structured data | Median weight |
|---|---|---|---|
| Antigua and Barbuda | 7 | 86% | 2.53 MB |
| Bahamas | 7 | 71% | 2.9 MB |
| Barbados | 7 | 57% | 0.87 MB |
| British Virgin Islands | 7 | 57% | 1.18 MB |
| Cayman Islands | 7 | 71% | 4.09 MB |
| Jamaica | 6 | 50% | 1.32 MB |
| Saint Lucia | 6 | 50% | 1.1 MB |
| Trinidad and Tobago | 5 | 20% | 1.87 MB |
Small per-territory sample sizes (5–7 sites) mean these splits are indicative, not precise.
| Sector | Sites measured | Has structured data | Median weight | Median TTFB |
|---|---|---|---|---|
| Professional and financial services | 15 | 53% | 0.93 MB | 0.59s |
| Hotels and villas | 12 | 75% | 3.96 MB | 0.56s |
| Construction and trades | 11 | 45% | 1.18 MB | 2.26s |
| Marine and charter | 9 | 67% | 1.86 MB | 0.33s |
| Restaurants | 4 | 50% | 2.11 MB | 0.8s |
| Retail | 1 | 100% | 8.79 MB | 0.52s |
The retail row covers a single site and should not be read as a sector finding.
The region has solved the visible problems. Every measured site was served over valid HTTPS and almost all were built to work on a phone. Those were the open questions a decade ago and they are now settled.
What has not been solved is machine readability. Two in five homepages tell a crawler nothing structured about the business, only one in ten declares itself a local business with an address and service area, and not a single site in the sample published FAQ markup. That matters more than it used to, because an AI assistant answering “who does X in Tortola?” is assembling an answer from entities it can parse, not from a page it has admired.
Weight and response time are the second gap. A median 1.84 MB homepage — a lower bound, since we counted only the first 40 assets — is a meaningful wait on a Caribbean mobile connection, and a third of sites had not sent their first byte after a full second. Hotel and villa sites carried the heaviest pages in the sample at a 3.96 MB median, which is precisely the sector whose visitors are browsing on holiday data.
The encouraging read is that these are cheap fixes relative to their effect. Structured data, a sitemap, a sensible meta description and disciplined image weight are days of work, not a rebuild — and on the evidence above, doing them puts a site ahead of most of its regional peers.
We assembled 55 publicly reachable business websites across eight English-speaking Caribbean territories and six sectors, then requested each homepage once with a mobile user agent on 31 July 2026, recording response time, transfer size, mobile viewport, title and meta description, JSON-LD structured data, robots.txt and sitemap.xml. 52 sites returned usable measurements; three failed on DNS or TLS errors. No Phoenix Caribbean client sites were included.
Structured data. 40% of the homepages we measured carried no JSON-LD at all, only 10% identified themselves as a local business with an address and service area, and none published FAQ markup. Basic transport hygiene is in far better shape: 100% of measured sites used valid HTTPS and 98% had a mobile viewport tag.
The median measured homepage weighed 1.84 MB counting the HTML plus its first 40 assets, and a quarter of sites exceeded 3 MB on that partial count — so real totals are higher. Hotel and villa sites were heaviest at a 3.96 MB median; professional and financial services sites were lightest at 0.93 MB.
Yes. The per-site measurements sit behind this summary and we are happy to share the figures for a specific territory or sector, or to run the same scan against your own site. Book a 30-minute consultation and we will walk through where your site sits against the sample.
We will run the same scan against your website and walk you through the results in a free 30-minute consultation.
