Direct answer: Of 41 cached article pages, 36 contain images, 514 <img> tags in total: 395 have non-empty alt (76.9%), 24 have empty alt and 95 have no alt. 19 pages cover every image, 17 cover part, none cover zero; the missing ones concentrate in E09, E16, E26 and E28. This is well above the homepage rate of 51.3%.

Study 38 looked at homepages: 51.3% of 978 images had alt. This study applies the same three-state rule to article pages to test whether “bad on the homepage” also means “bad inside”.

41
cached article pages
36
with images
514
images
395
with non-empty alt
95
missing alt
76.9%
non-empty alt share

Method: the same three states

Pages come from the 41 successful crawls in study 29, with raw HTML stored off-repository. Every <img> is classified as non-empty alt, empty alt (alt="" or whitespace) or missing alt. Text quality is not judged and JavaScript is not executed.

StateRuleMeaning
Non-empty altalt exists and is not blankAt least some alternative text
Empty altalt="" or whitespaceMay be decorative, may be an omission
Missing altNo alt attributeNo declaration at all

Result: 76.9% have alt

ItemCountShare
Images514100%
Non-empty alt39576.9%
Empty alt244.7%
Missing alt9518.5%
Page coveragePagesNote
All images have alt19E03, E06, E09, E14, E23 and others
Partial17Still has missing or empty alt
No alt at all0No page with zero coverage
No images5Excluded from the rate

The median non-empty share among pages with images is 100%: more than half cover every image. That differs from the homepage split, which was closer to “all or nothing”.

Against homepages: 76.9% vs 51.3%

LayerPages / imagesNon-empty altEmpty altMissing alt
Homepages (study 38)25 / 97851.3%24.6%24.0%
Article pages (this study)41 / 51476.9%4.7%18.5%

Article pages cover better, but the two groups differ in page type and selection, so this is not a controlled comparison. A plausible explanation is that article templates carry fewer images and name them more carefully, while homepages include many icons, decorative graphics and carousel slides. Without per-image purpose labels, that cannot be confirmed.

Missing alt concentrates on a few pages

The 95 missing alt tags come mainly from E09 (15), E16 (11 on each of three pages), E26 (11) and E28 (9). These pages often carry thumbnail lists or older templates. At the other end, E03's 39 images and E14's 25–27 images are fully covered.

Alt coverage is a template-level difference: pages from the same site behave alike, and the gap between sites is larger. Acceptance should sample by template or site rather than a random single page.

What this means for acceptance

  1. Inner pages are not necessarily worse. Article pages scored 76.9% against 51.3% on homepages.
  2. Sample by template. Pages from the same template behave similarly; cross-template gaps are larger.
  3. Missing alt is concentrated. Most of the 95 come from four sites.
  4. Keep empty and missing alt apart. Article pages had only 24 empty but 95 missing.
  5. Text quality is a separate test. Having alt is not the same as describing accurately.

Limits

  • Cached, zero-request: pages are frozen at the study-29 crawl.
  • Attributes only: no judgement of accuracy, duplication or keyword stuffing.
  • No JavaScript: lazy-loaded and carousel images are invisible.
  • Mixed page types: the article sample may include list pages with different image mixes.
  • 41-page pilot: no industry claim.

Data and reproduction

Anonymised data: equipment-article-image-alt-2026.csv (41 rows: sample_id, URL, images, non-empty alt, empty alt, missing alt, share). The repository keeps docs/research/2026-11-article-image-alt.csv.

Parser: scripts/img_alt_probe.py; unit tests: scripts/test_img_alt_probe.py (10 tests, no network). Run:

python -m unittest scripts.test_img_alt_probe

Related: the homepage image alt pilot (978 images), the expanded article-page dates study, the contact message-form pilot, and the homepage image attributes pilot (dimensions 1.7%, lazy 0).

Sources and definitions

  1. Page cache: 41 successful pages from study 29; raw HTML kept off-repository.
  2. Alt states: non-empty / empty / missing; static HTML only, no JavaScript.
  3. Scan: 2026-11-03, zero network requests.
About the author

Lingyi Information Official (delivery team) works on requirements, proposals and acceptance for corporate websites, brand and smart-manufacturing projects. This article counts cached image attributes only; it does not judge alt-text quality and makes no claim about any site's rankings or traffic; the 41-page sample is a pilot.

Inner pages are not necessarily worse

Sample by template and keep non-empty, empty and missing alt apart; read coverage per site and test text quality separately.

Request a recheck plan