Direct answer: Of 61 cached pages, only 17 declare og:image (27.9%) and 44 have no share image at all; none of the 25 homepages has one. og:title appears on 39 pages, og:description on 39 and og:url on 38. Twitter Card appears on 38 pages, all summary_large_image, but twitter:image is 0; 21 of those pages have no og:image either.
All 17 og:image values are absolute, but none declares og:image:width or height: the image exists, yet platforms still have to crop it themselves.
Method: read meta from two caches
The sample is 25 homepage caches (2026-10-31 to 11-01) plus 36 article pages from 10 sites (fetched 2026-11-03 08:40–08:41; five JS shells excluded), 61 pages in total and zero network requests.
| Item | Rule | Boundary |
|---|---|---|
| OG fields | meta property: title / description / url / image | Both attribute orders handled |
| Image size | og:image:width / height present | Image not fetched; real size not checked |
| Twitter Card | meta name: card / title / image | Type and presence only |
| Has share image | og:image or twitter:image present | Which image a platform picks is unknown |
| Large card without image | summary_large_image and empty twitter:image | Platforms may fall back to og:image |
Result: titles are covered, images are not
| Declaration | Pages | Share of 61 |
|---|---|---|
| og:title | 39 | 63.9% |
| og:description | 39 | 63.9% |
| og:url | 38 | 62.3% |
| og:image | 17 | 27.9% |
| og:image:width / height | 0 | 0% |
| twitter:card | 38 | 62.3% |
| twitter:title | 36 | 59.0% |
| twitter:image | 0 | 0% |
Of the 39 pages with og:title, 22 have no og:image: a share may show only text, without a thumbnail.
Homepages vs articles: the image gap is on homepages
| Page type | Pages | og:title | twitter:card | og:image | No image at all |
|---|---|---|---|---|---|
| Homepage | 25 | 9 | 8 | 0 | 25 |
| Article | 36 | 30 | 30 | 17 | 19 |
| Total | 61 | 39 | 38 | 17 | 44 |
All 25 homepages lack og:image and only 9 have og:title; articles have 17 og:image values, but 19 still have no share image at all.
38 large cards, 0 twitter:image
| Item | Count | Note |
|---|---|---|
| summary_large_image | 38 | Every card is this type |
| With empty twitter:image | 38 | 100% |
| Also without og:image | 21 | No image to fall back to |
| Pages with twitter:image | 0 | — |
A “large image” card without twitter:image may fall back to og:image, but 21 pages have neither, so only a text card or a default site image can be shown.
What this means for acceptance
- Add og:image to homepages first. All 25 lack it, so a shared homepage has no thumbnail.
- Check title and image together. 22 of the 39 og:title pages have no image.
- Pair large cards with twitter:image. All 38 large cards lack their own image.
- Declare width and height. None of the 17 images does, so platforms cannot predict the ratio.
- Nineteen article pages still lack images. The article template does not apply the field consistently.
Limits
- Meta only: images are not fetched and size, format, availability and actual previews are not tested.
- Platform fallback unknown: WeChat, LinkedIn, Slack or X may use the first body image, a JSON-LD image or a site icon.
- Copy not judged: whether og:title/description match the page or are too long is not measured.
- Sample shape: 61 pages from 25 homepages and articles on 10 sites; the two caches differ in date.
- 25-site pilot: no industry claim.
Data and reproduction
Anonymised data: equipment-social-cards-2026.csv (61 rows × 15 columns: page type, OG fields, absolute image flag, dimensions, Twitter Card, image source and the large-card-without-image flag). The repository keeps docs/research/2026-11-social-cards.csv.
Parser: scripts/social_card_probe.py; unit tests: scripts/test_social_card_probe.py (10 tests, no network). Run:
python -m unittest scripts.test_social_card_probe
Related: the article head metadata pilot, the mobile viewport pilot, and the unknown path and soft 404 pilot., and the article heading outline pilot
Sources and definitions
- Homepage cache: 25 homepages from study 4 (2026-10-31 to 11-01).
- Article cache: 36 pages from 10 sites in study 45 (2026-11-03 08:40–08:41), five JS shells excluded.
- Rules: meta property/name string matching; absolute image by URL prefix.
- Scan: 2026-11-03, zero network requests.