Direct answer: A second-pass parse of the same 20 contact pages: address 19/20, phone 19/20, email 16/20 — basic contact details are close to standard. Operational details are not: business hours 0/20, response promise 2/20, a named department 3/20; WeChat 8/20, map 13/20, privacy notice 5/20. Landlines 14, mobiles 12, 400 numbers only 4. This is a pilot, not the industry.
The earlier pilot asked whether a homepage offers a contact entry and whether the page has a form. This one asks a different question: once you are there, what does the page actually tell you? No new requests — it parses the 20 contact-page HTML files cached on 2026-11-01 between 18:41 and 18:47.
Method: second pass, no new requests
The sample matches the earlier contact pilot: 30 companies → 25 analysable homepages → 21 contact entries → 20 contact pages returning HTTP 200. This round reads only the saved HTML, applies text-feature rules, and outputs 20 anonymised rows.
| Field | Rule (text feature) |
|---|---|
| Address | "地址 / Address / Add:" appears |
| Phone | A landline, mobile or 400 number pattern appears |
| An email address appears | |
| WeChat / map | "微信/二维码/扫码" or "地图/高德/百度地图" appears |
| Business hours | "营业时间/工作时间/上班时间/周一至" appears |
| Response promise | "24 小时内 / 一个工作日 / 尽快回复" appears |
| Named contact / department | "联系人/经理/先生/女士" or "销售部/采购部/外贸部" |
| Privacy notice | "隐私 / 个人信息 / privacy" |
| Form fields | Count of input/textarea/select inside <form> |
A match does not prove the detail is accurate or current; a miss does not prove absence (it may sit in an image, PDF, JS pop-up or QR code). Numbers are checked by format only, not by dialling.
Result 1: basic details are nearly standard
| Detail | Present | Share of 20 |
|---|---|---|
| Address | 19 | 95% |
| Phone (any format) | 19 | 95% |
| 16 | 80% | |
| Map | 13 | 65% |
| WeChat / QR code | 8 | 40% |
Address and phone coverage (95%) is higher than the earlier pilot's "recognisable inquiry form" (9/20). That explains a common misread: the page can look complete while its lead capture is not. Having an address and a phone is not the same as having a working form.
Result 2: phone formats
| Number type | Present | Note |
|---|---|---|
| Landline (0xx-xxxxxxx) | 14 / 20 | Usually a switchboard or sales office |
| Mobile (1[3-9]xxxxxxxxx) | 12 / 20 | Usually a salesperson's direct line |
| 400 hotline | 4 / 20 | Clearly rarer than landline and mobile |
| At least one | 19 / 20 | One page offers only email/forms |
Mobiles and landlines coexist (12 and 14), so the "personal direct line" remains the norm here and a 400 line is not the default. For a buyer this means the experience depends on whether one salesperson is at their desk, not on a single front door.
Result 3: operational details are the gap
| Operational detail | Present | Share of 20 |
|---|---|---|
| Business / working hours | 0 | 0% |
| Response-time promise | 2 | 10% |
| Named department (sales, procurement, export) | 3 | 15% |
| Named person or title | 8 | 40% |
| Privacy notice | 5 | 25% |
0/20 means "not found in text", not "the company has no hours". One page mentions a working day and none states actual opening hours. For a buyer in another time zone, or after 6 pm, "when can I reach someone" is still a guess.
Response promises (2/20) and departments (3/20) are similarly thin. Together: the page answers "how do I contact you" but not "when, who, and how fast".
Result 4: form field counts
Fifteen pages contain a form, with a median of 7 fields (0–76). The count includes search and newsletter inputs, so it is not the number of inquiry fields: the earlier rule (email/tel/textarea plus submit/text) found only 9/20. A long form can be a site search or a full qualification form — count alone says nothing about quality.
Five actionable conclusions
- Address and phone are not a differentiator. 95% have them; the difference is form usability and operational detail.
- State your business hours. None of the 20 does; adding hours, time zone and holidays is cheap.
- Give a response promise. "Within one business day" beats "welcome to contact us".
- Name a department or role. Routing procurement, technical and after-sales saves a transfer.
- Put the privacy notice next to the form. 25%, unchanged from the earlier pilot; still the weakest item.
Limits
- Text features: mixed Chinese/English, traditional characters and text in images are missed; results may undercount.
- Static view only: PDFs, JS pop-ups and QR-code content are out of scope.
- Numbers unchecked: format only; no dialling, no owner verification.
- Non-random sample: one list, one city, 20 pages; not the industry.
- Not a recheck: capture time is still 2026-11-01 18:41–18:47; no site was revisited.
Data and recalculation
Anonymised data: equipment-contact-details-2026.csv (20 rows, 17 fields). The repository keeps docs/research/2026-11-contact-details.csv.
Parser: scripts/contact_details_probe.py; tests: scripts/test_contact_details_probe.py (16 tests, no network). Run:
python -m unittest scripts.test_contact_details_probe
Related: the 25-homepage / 20-contact-page pilot, the 24-site four-layer review and the 12 acceptance clauses (item 7).
Sources and definitions
- Frame: 2024 Changzhou first-set major equipment list, positions 71–100; contact pages captured 2026-11-01 18:41–18:47 Asia/Shanghai.
- ContactPage vocabulary: schema.org.
- Zero requests this round; raw HTML stays outside the repository.