Direct answer: only 17 of 25 homepages (68%) offer at least one contact path, and eight pages have no form, tel, mailto or contact-page link at all. A message form appears on four homepages and site search on six; phone links appear on 11, contact-page links on 15 and email links on seven.
The sample holds 22 forms and 54 fields, 32 of which (59.3%) have an accessible label. Path counts split as eight pages with none, six with one, four with two, five with three and two with all four. On the 10 sites present in both samples, homepage and article pages carry exactly the same kinds of path (10 of 10); the overall ratio gap comes from sample composition, not from the page type.
Method: reuse the study-4 homepage cache and the study-68 rules
The study reads the 25 homepage caches captured on 2026-10-31 with zero network requests and submits nothing. The parser is the same module used for the study-68 article pages: form counts, message and search forms, fields and accessible labels, tel / mailto / contact links, QR and WeChat wording, and a 0–4 path score.
| Item | Rule | Boundary |
|---|---|---|
| Message form | Form contains a textarea | JS-inserted forms excluded |
| Search form | type=search, name=q/keyword or action containing search | Otherwise unclassified |
| Fields | input/select/textarea except hidden and buttons | Visibility not judged |
| Accessible label | aria-label, aria-labelledby, title, placeholder, label for or wrapping label | Wording not judged |
| Contact link | href containing contact / lianxi / lxwm / about / feedback | Keyword lower bound |
| Path score | How many of message form, tel, mailto, contact link exist | 0–4, QR/WeChat excluded |
Result: nearly a third of homepages have no contact path
| Path | Pages | Share of 25 |
|---|---|---|
| Contact-page link | 15 | 60.0% |
Phone link tel: | 11 | 44.0% |
Email link mailto: | 7 | 28.0% |
| WeChat wording | 7 | 28.0% |
| Site search form | 6 | 24.0% |
| QR wording | 6 | 24.0% |
| Message form | 4 | 16.0% |
| At least one path | 17 | 68.0% |
Counts: eight pages with no path, six with one, four with two, five with three and two with all four. Only E02 and E17 reach four; E01, E12, E18, E20, E25, E27, E28 and E29 have no usable contact path in the HTML at all.
Forms and fields: 22 forms, 54 fields
| Type | Count | Note |
|---|---|---|
| Forms | 22 | Across 14 homepages |
| Message forms (textarea) | 4 | Two on E02, one each on E11 and E17 |
| Identifiable search | 6 | One per page on six pages |
| Fields | 54 | Up to 22 on one page (E02) |
| Labelled fields | 32 | 59.3% |
E02 carries four forms, 22 fields and nine labels, the densest page here; E26's six fields all lack a programmatic label (the same site and template as its article-page form in study 68). E13 and E28 each have a search form with a labelled field, while E01, E04 and E29 have single-field forms that do not map to any contact path.
By site
| Site | Forms | Message | Search | tel | mailto | Contact | Paths |
|---|---|---|---|---|---|---|---|
| E01 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
| E02 | 4 | 2 | 0 | 2 | 1 | 10 | 4 |
| E03 | 2 | 0 | 0 | 9 | 1 | 7 | 3 |
| E04 | 1 | 0 | 0 | 0 | 0 | 13 | 1 |
| E06 | 1 | 0 | 0 | 14 | 0 | 21 | 2 |
| E08 | 2 | 0 | 2 | 3 | 0 | 5 | 2 |
| E09 | 0 | 0 | 0 | 2 | 1 | 5 | 3 |
| E11 | 2 | 1 | 1 | 0 | 0 | 18 | 2 |
| E12 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| E13 | 1 | 0 | 1 | 0 | 0 | 10 | 1 |
| E14 | 0 | 0 | 0 | 4 | 1 | 0 | 2 |
| E16 | 0 | 0 | 0 | 1 | 0 | 0 | 1 |
| E17 | 2 | 1 | 1 | 1 | 5 | 12 | 4 |
| E18 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| E19 | 1 | 0 | 1 | 0 | 0 | 7 | 1 |
| E20 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| E22 | 0 | 0 | 0 | 3 | 2 | 6 | 3 |
| E23 | 2 | 0 | 0 | 3 | 2 | 18 | 3 |
| E24 | 0 | 0 | 0 | 0 | 0 | 1 | 1 |
| E25 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| E26 | 1 | 1 | 0 | 1 | 0 | 24 | 3 |
| E27 | 0 | 0 | 0 | 0 | 0 | 0 | 0 |
| E28 | 1 | 0 | 1 | 0 | 0 | 0 | 0 |
| E29 | 1 | 0 | 0 | 0 | 0 | 0 | 0 |
| E30 | 0 | 0 | 0 | 0 | 0 | 52 | 1 |
E30 has 52 contact-page links but no tel, mailto or form; E28 and E29 have a form yet no identifiable path, because their single text input is neither search nor message. E04 carries 13 contact links and one form but no tel or mailto.
Compared with article pages
Updated in place on 2026-11-05 20:45: added a 10-site paired check and corrected the earlier page-type reading of the totals.
| Measure | Homepages (this study, 25) | Articles (study 68, 36) |
|---|---|---|
| At least one path | 17 (68.0%) | 35 (97.2%) |
| No path at all | 8 (32.0%) | 1 (2.8%) |
| Message form | 4 (16.0%) | 1 (2.8%) |
| Phone link | 11 (44.0%) | 34 (94.4%) |
| Email link | 7 (28.0%) | 25 (69.4%) |
| Contact-page link | 15 (60.0%) | 27 (75.0%) |
| Labelled fields | 32/54 (59.3%) | 26/32 (81.2%) |
Paired check: 10 sites with both page types
The study-4 homepage sample and the study-45 article sample overlap on 10 sites. Pairing them by site shows that homepage and article pages carry exactly the same kinds of path:
| Site | Homepage paths | Article pages | Article paths (per page) | Match | Message form (home / article) |
|---|---|---|---|---|---|
| E03 | 3 | 5 | 3 | yes | 0 / 0 |
| E06 | 2 | 5 | 2 | yes | 0 / 0 |
| E09 | 3 | 5 | 3 | yes | 0 / 0 |
| E14 | 2 | 5 | 2 | yes | 0 / 0 |
| E16 | 1 | 3 | 1 | yes | 0 / 0 |
| E19 | 1 | 1 | 1 | yes | 0 / 0 |
| E22 | 3 | 5 | 3 | yes | 0 / 0 |
| E23 | 3 | 5 | 3 | yes | 0 / 0 |
| E26 | 3 | 1 | 3 | yes | 1 / 1 |
| E28 | 0 | 1 | 0 | yes | 0 / 0 |
All 10 sites match: sites with tel on the homepage also have tel on their article pages, E28 has no path on either, and E26 is the only site with a message form — one on the homepage and one on its article page. Only the counts differ: E03 has nine tel: links on its homepage but one per article page.
Path kinds are therefore driven by the site template, not by the page type. The 68% and 97.2% totals differ because of sample composition: 8 of the other 15 homepage-only sites have no path at all.
The two samples cover different site sets (25 homepages on 2026-10-31; 10 sites' articles on 2026-11-03). The paired check shows that the 10 shared sites have identical path kinds (10 of 10), so the totals above cannot be read as a page-type effect; they only show that this homepage sample contains more sites with no path. The two snapshots are three days apart and are not a redesign comparison.
What this means for acceptance
- Check the "does a path exist" bar first. Eight of 25 homepages have no form, tel, mailto or contact link in HTML; a mobile visitor may have to hunt through the menu.
- Put message forms on homepages and articles. Four of 25 and one of 36, far below contact pages (15 of 20).
- A phone link beats phone text. Only 11 homepages use
tel:; the rest show a number that cannot be tapped to call. - Label form fields programmatically. 22 of 54 homepage fields have no label; E26's six fields rely on visible text only.
- Do not let a QR code replace an on-site path. Six homepages mention one, but scanning leaves no on-site record.
Limits
- HTML only: JS-inserted floating widgets, pop-ups and forms are invisible, so counts may be low.
- Page level: the same form repeated in header and footer is counted twice.
- Keyword lower bounds: contact links match on href, QR/WeChat on text and alt.
- Different samples: 25 homepages and 10 sites' articles are not the same sites; no causal or before/after claim.
- 25 sites: a pilot, no industry claim.
Data and reproduction
Anonymised data: equipment-homepage-inquiry-paths-2026.csv (25 rows × 13 columns: forms, message forms, search forms, fields, labelled fields, tel/mailto/contact links, QR and WeChat, path count). The repository keeps docs/research/2026-11-homepage-inquiry-paths.csv.
Parser: scripts/article_inquiry_probe.py (the same implementation as study 68); unit tests: scripts/test_article_inquiry_probe.py (no network, nine tests). Run:
python -m unittest scripts.test_article_inquiry_probe
Related: the article inquiry-path pilot, the inquiry path checker, and the homepage form-input pilot.
Sources and definitions
- Homepage cache: 25 homepages from study 4 (2026-10-31).
- Forms: input/select/textarea inside
<form>, excluding hidden fields and buttons. - Scan: 2026-11-05, zero network requests, no form submitted.