Files
jobboard/docs/plan-hosts/own-1.md
T

61 lines
6.4 KiB
Markdown

# Own-site batch 1: how each career site can be read
31 companies, all classified from live responses (2026-09-29).
## Counts per class
| class | n | companies |
|---|---|---|
| known-host | 9 | ada-health (greenhouse, 0 roles), dfds (teamtailor x11 subdomains), lunar (teamtailor), idealista (teamtailor), checkout-com (ashby), skyscanner (ashby), tomtom (lever EU), whereby (lever, 0 roles), otovo (recruitee) |
| new-host | 8 | pandora, terma, universal-robots, volvo-cars (all SAP SuccessFactors), siemens (avature), microsoft (eightfold), king (phenom), sinch (oracle recruiting cloud) |
| own-api | 4 | dynatrace (Coveo proxy), new-work (Algolia), relex-solutions (WordPress REST), vinted (Next.js data with Greenhouse records) |
| html | 5 | check24, zalando, bolt, lucca, bird |
| jsonld | 2 | sophia-genetics, aiven |
| browser | 1 | meta |
| none | 2 | wefox, unbabel |
## Counts per host
- successfactors: 4 (pandora `pandoraas`, terma `termaas`, universal-robots via Teradyne `teradynein`, volvo-cars `C0000870892P`)
- teamtailor: 3 clean (lunar, idealista, dfds with 11 country boards); unbabel has a dangling Teamtailor custom domain
- ashbyhq: 2 (checkout-com slug `checkout.com`, skyscanner slug is a UUID)
- lever: 2 (tomtom on the EU instance, whereby)
- greenhouse: 1 clean (ada-health, empty); vinted and aiven show Greenhouse traces but no board token was found
- avature, eightfold, phenom, oracle recruiting cloud: 1 each
- recruitee: 1 (otovo)
Discovery misses worth fixing in the app: Lever EU (`api.eu.lever.co`), custom domains for Teamtailor and Recruitee (`careers.<domain>`), an Ashby slug that contains a dot, an Ashby slug that is a UUID visible only in a JS chunk, and Greenhouse boards linked from a button rather than from the page source.
## New hosts and how the listing is read
**SAP SuccessFactors (career5 / career2 / career41.sapsf.com).** All four companies expose the standard public XML feed `https://<careerN host>/career?company=<id>&career_ns=job_listing_summary&resultType=XML`, which returns every requisition as a `<Job>` element with title, description, req id and dates in one response (0.3 to 5 MB). Verified: pandora 583 roles, terma 104, volvo-cars 40 (matches its sitemap and the "of 40" search text), teradyne 486 in total. The company id comes from the `company=` parameter in the apply or register links on the career page (`jobs.volvocars.com`, `jobs.teradyne.com` and the Terma page carry it). For Universal Robots the feed is the whole Teradyne group; UR roles are the 30 sitemap URLs under `/Universal-Robots/job/` of `jobs.teradyne.com`, so a filter on that path or the business-unit field is needed. Pandora's own site is a Paradox.ai template over the same data.
**Avature (jobs.siemens.com, portal 144).** The list page `/en_US/externaljobs/SearchJobs` is server-rendered but shows 6 rows per page and "999+ results". An RSS feed at `/en_US/externaljobs/SearchJobs/feed/` returns 20 `<item>` elements with job links; `jobOffset` and `jobRecordsPerPage` did not change the result, so the feed alone cannot enumerate Siemens' roles. The sitemap lists only static pages. Job pages have no JobPosting JSON-LD. A reader would have to page the HTML search or use filter URLs; not solved.
**Eightfold (apply.careers.microsoft.com).** `GET /api/pcsx/search?domain=microsoft.com&query=&start=0&num=10` returns JSON with `data.positions` and `data.count` (2319 seen). It needs no login, but the first call returned 429 and other paths (`/api/apply/v2/jobs`) return 403 "Not authorized for PCSX", so it must be polled slowly with a normal browser User-Agent. Paging is by `start` and `num`. Location filters go in extra query parameters.
**Phenom (careers.king.com).** POST `https://careers.king.com/widgets` with the JSON body the site's own script sends (`ddoKey: refineSearch`, `pageName: search-results`, `size`, `from`, `jobs: true, counts: true`, `country: us`, `lang: en_us`) returns `refineSearch.totalHits` (33) and the job objects. The sitemap `/sitemap.xml` also lists the 33 `/job/` URLs. `refNum` is `ACMKDSUS`.
**Oracle Recruiting Cloud (iaings.fa.ocs.oraclecloud.com, site SinchCareer).** `GET /hcmRestApi/resources/latest/recruitingCEJobRequisitions?onlyData=true&finder=findReqs;siteNumber=CX_1,limit=25,offset=0` returns JSON with `TotalJobsCount` (79) and a `requisitionList`; page with `offset`. The site number `CX_1` is on the candidate-experience page as `data-sitenumber`. Sinch's own pages give no link to this site; it was found by web search, and Sinch also has an empty Workable account.
## Own-site notes
- dynatrace: `POST /api/coveo/search/` with `{"q":"","numberOfResults":1000}` gives `totalCount` 108 and the results; the sitemap and JobPosting JSON-LD are a fallback.
- new-work: the given `new-work.se/en/career` is a stub; `corporate.xing.com/en/career` ships `window.algoliaSettings` (public search key, app `4TACNCQ35A`, index `new-work-se-live-en`). The query only works with a `Referer` of the corporate site; 17 job hits with `filters: type:Job`.
- relex-solutions: standard WordPress REST `wp-json/wp/v2/job`, 43 items (`X-WP-Total`).
- vinted: `careers.vinted.com/jobs` embeds 201 Greenhouse-EU job records (`internal_job_id`, `gh_jid`) in an 8.9 MB page; the board token was not found (`boards-api.greenhouse.io` 404 for four guesses).
- dfds: `/api/trpc/careers.getPostings` aggregates 158 postings whose links point to per-country Teamtailor boards.
- tomtom, checkout-com: the site proxies Lever EU and Ashby respectively; the vendor APIs work directly.
## Not classified cleanly
- wefox: Wix site, careers under `careers-archiv`, no role list in the HTML, JOIN board 404. `none` with the caveat that a Wix-rendered widget could not be ruled out.
- unbabel: careers domain CNAMEs to Teamtailor but does not serve (TLS mismatch, 421), the main site moved to TransPerfect infrastructure, the Workable account is empty.
- meta: every careers HTML route returns a 400 error page to curl; the sitemap has no role URLs. `jobs_url` is the well-known `/jobsearch`, not verified by response.
- idealista: `www.idealista.com` is behind DataDome (captcha page); `careers.idealista.com` was tried directly and is Teamtailor, but the link from the main site could not be seen.
- sinch: see above; the connection to the Oracle site is by web search only.
## Method notes
One request per second per host, `curl -m 20 -L`, browser User-Agent, no login. The Algolia call for new-work used a Referer header matching the site that ships the key.