Direct answer: Of 25 cached homepages, only 3 use <main>, 4 use <nav>, 4 use <header> and 3 use <footer>; <article> and <aside> appear nowhere. 12 homepages have no H1, 8 have exactly one and 5 have several (up to 14); 18 have no H2, and 8 pages have no H1–H6 at all. Semantic structure is not the default on these homepages.
Earlier studies covered dates, alt text and privacy; this one looks at the underlying HTML: landmark elements and heading levels. They affect screen-reader navigation and how search and AI systems read page structure, and can be counted on cached HTML with zero requests.
Method: tag counts plus level flags
Using the 25 cached homepages from study 4, the study counts main / article / nav / header / footer / section / aside and h1–h6, then flags: no H1, multiple H1, no H2, H2 without H1, H3 without H2 and no headings at all. Static HTML only; no JavaScript and no ARIA inspection.
| Item | Rule | Boundary |
|---|---|---|
| Landmark | Count an opening tag when present | Visibility and duplication not checked |
| Heading | Count h1–h6 tags | Empty text not checked |
| Skipped level | H2 without H1; H3 without H2 | Adjacent levels only |
| No headings | All of h1–h6 are zero | Div/span headings are invisible to this method |
Landmarks: only 3 use main
| Tag | Sites (25) | Note |
|---|---|---|
<main> | 3 | E02, E17, E30 |
<nav> | 4 | E02, E17, E24, E25 |
<header> | 4 | E02, E19, E24, E28 |
<footer> | 3 | E02, E19, E28 |
<section> | 3 | E16, E17, E24 |
<article> | 0 | — |
<aside> | 0 | — |
Headings: 12 without H1, 5 with several
| H1 count | Sites | Note |
|---|---|---|
| 0 | 12 | E01, E03, E09, E12, E16, E18, E20, E25, E26, E27, E29, E30 |
| 1 | 8 | Structurally clearest |
| 2–14 | 5 | E11 (14), E17 (2), E22 (3), E24 (9), E28 (6) |
Multiple H1s usually come from templates that treat section and card titles as H1; on the page with 14, H1 has lost its role as the page title. Eighteen sites have no H2 at all, so heading structure is either one level or absent.
Skipped levels and pages without headings
- H2 without H1: 2 sites (E03, E30).
- H3 without H2: 5 sites (E01, E19, E22, E24, E28).
- No H1–H6 at all: 8 sites (E12, E16, E18, E20, E25, E26, E27, E29).
Pages without headings may simulate them with bold text inside a <div>. They can look fine but carry no machine-readable hierarchy. For acceptance, that is a structural gap, not a styling issue.
What this means for acceptance
- Check whether an H1 exists. Twelve of 25 have none; it is not a default.
- There should be exactly one H1. Five sites have several, up to 14.
- Missing H2 is common. Eighteen sites have none, so sections may be styled only.
- Landmarks are a bonus, not the norm. Only 3 use main and none use article.
- List the no-heading pages. Eight sites need this fixed first.
Limits
- Cached, zero-request: pages are frozen at 2026-10-31 to 11-01.
- Tag counts: empty, hidden or duplicated headings are not judged.
- No JavaScript: script-generated headings and landmarks are invisible.
- No ARIA: roles and aria-label are out of scope.
- 25-site pilot: no industry claim.
Data and reproduction
Anonymised data: equipment-semantic-structure-2026.csv (25 rows: landmark counts, h1–h6 counts, no-H1/multi-H1/no-H2/skipped-level/no-heading flags). The repository keeps docs/research/2026-11-semantic-structure.csv.
Parser: scripts/semantic_structure_probe.py; unit tests: scripts/test_semantic_structure_probe.py (9 tests, no network). Run:
python -m unittest scripts.test_semantic_structure_probe
Related: the article date-to-H1 distance pilot (article H1), the homepage image alt coverage and text study, and the homepage privacy and analytics pilot.
Sources and definitions
- Homepage cache: 25 homepages from study 4 (2026-10-31 to 11-01).
- Structure: tag counts and adjacent-level flags; no JavaScript, no ARIA.
- Scan: 2026-11-03, zero network requests.