# Taleo Jobs API > Query jobs from Taleo boards through one API with ?source=taleo — plus how Taleo's sharded searchResults endpoint really behaves. Source: https://hyperjobs.io/ats/taleo Taleo is the ATS Oracle bought in 2012, and the name covers **two different products** that share a brand and almost nothing else. We cover one of them: Taleo Business Edition, the SMB and mid-market tier that lives on `tbe.taleo.net`. It's in this dataset at 712 postings, 710 of them carrying a description, which are numbers that move hourly, so read them from [`/v1/meta`](/docs/api-reference/meta) rather than trusting this sentence. We do not cover Taleo Enterprise. That asterisk leads this page rather than sitting in a footnote, because "we support Taleo" means nothing until you know which Taleo the employer you want is running. ## Which Taleo is it | | Taleo Business Edition (TBE) | Taleo Enterprise Edition (TEE) | | --- | --- | --- | | Board URL | `.tbe.taleo.net//ats/careers/v2/…` | `.taleo.net/careersection/…` | | Sells to | SMB and mid-market | Large enterprise | | In this dataset | **Yes** — `?source=taleo` | **No** — see below | The board URL is the whole check. If the employer you want serves `careersection`, nothing on this page covers them, and the fix is usually [`?source=oracle`](/ats/oracle) rather than anything Taleo-branded. ### Why Enterprise isn't the target TEE is a stateful JSF application, and it doesn't behave like a site that happens to be scrapeable. Its `jobsearch.ftl` returns a healthy `200` containing **zero** job links: the shell renders, and the jobs arrive later through application state a plain client isn't carrying. The REST endpoint the competitor guides document needs a per-tenant `portal` id discovered from that tenant's own `jobsearch.ftl`, so it isn't enumerable. Those guides' own worked example, `hdr.taleo.net`, now returns a `500`. The one portal id we did find turned out to be a template default shared across four unrelated tenants, which is worse than no id, because it looks like a discovery. A `403`, a challenge page or a CAPTCHA means the server decided you're a robot, and the answer is a better TLS fingerprint or a residential IP. TEE does none of that. It returns a real `200` and a real page that genuinely has no jobs in it yet. Nothing is refusing us. There is simply nothing to parse until an application runs. So impersonation tiers and residential proxies are the **wrong tools** here, and buying more of them would spend money to receive the same empty shell faster. TEE needs a browser. That's a different tier, and it isn't built. There's a second reason not to fight it: most of the TEE base has **migrated to Oracle Cloud HCM**, which [`?source=oracle`](/ats/oracle) already covers and which is one of the largest sources in this dataset. The Taleo Enterprise tenants worth having are mostly Oracle Cloud tenants now, so TBE plus Oracle Cloud spans the reachable Taleo universe and standalone TEE is skipped on purpose rather than by accident. ## Try it The value is `taleo`, and it means Business Edition. `?source=taleo` returns TBE rows and only TBE rows. It is not an error and not a gap in the crawl if a `careersection` employer you expected isn't among them. ```bash curl "…/v1/jobs?source=taleo" # ✓ 712 postings, all Business Edition ``` ```bash curl -G https://api.hyperjobs.io/v1/jobs \ -H "Authorization: Bearer $HYPERJOBS_KEY" \ --data-urlencode "source=taleo" \ --data-urlencode "time_frame=30d" \ --data-urlencode "limit=2" ``` ```json title="Illustrative — the shape, not a live response" { "data": [ { "id": "taleo:NORTHWIND2_4137", "title": "Staff Accountant", "company": { "slug": "northwind-associates", "name": "Northwind Associates", "domain": "northwind.com", "website": "https://northwind.com", "industry": "Financial Services", "employee_count": 340 }, "location": { "remote": false, "countries": ["United States"], "cities": ["Wakefield"], "regions": ["Massachusetts"], "raw": ["Wakefield, MA, Massachusetts, US"] }, "employment_type": ["FULL_TIME"], "work_arrangement": null, "seniority": null, "salary": null, "taxonomies": ["Finance"], "apply_url": "https://phh.tbe.taleo.net/phh02/ats/careers/v2/viewRequisition?org=NORTHWIND2&cws=1&rid=4137", "url": "https://phh.tbe.taleo.net/phh02/ats/careers/v2/viewRequisition?org=NORTHWIND2&cws=1&rid=4137", "source": "taleo", "posted_at": "2026-06-26T00:00:00.000Z", "expires_at": null, "updated_at": "2026-07-14T11:20:03.000Z" } ], "total": 2, "limit": 2, "offset": 0 } ``` Three shape notes, all of them TBE telling you less than a modern board does. `apply_url` and `url` are identical, because TBE publishes no distinct apply link in what we read and `apply_url` [falls back to `url`](/docs/objects/job) — not true of [Lever](/ats/lever). `posted_at` is always midnight, because TBE dates arrive day-precision as `2026-06-25 00:00:00.0`; the date is real, the time isn't, on every Taleo row. And `work_arrangement` is usually `null`: the JSON-LD carries `jobLocationType` only when a job is remote, so a TBE board can tell you a role *is* remote but never that it isn't. We leave that null rather than reading the silence as on-site. ## How TBE boards actually work Server-rendered HTML on a sharded host, enumerable by URL pattern, with a list and a detail: ```bash # 1. List — page 0 is a fixed 10-job preview curl "https://.tbe.taleo.net//ats/careers/v2/searchResults?org=&cws=1" # 2. Pages 1..N — same session only curl "https://.tbe.taleo.net//ats/careers/v2/searchResults?next&rowFrom=10&org=&cws=1" # 3. Detail — schema.org JobPosting JSON-LD on the modern theme curl "https://.tbe.taleo.net//ats/careers/v2/viewRequisition?org=&cws=1&rid=" ``` Neither of those first two lines does what it looks like it does, and both fail quietly. ### Stale shards TBE boards are sharded across instances, and organisations get migrated between them: `phe01` becomes `phh02`. The URL on file still works, because it `302`s to the new shard. So you follow that redirect **manually** and read two things off it: the final base, since every later URL has to use the shard you landed on rather than the one you started from, and the `JSESSIONID` the redirect target sets, which the next section is entirely about. Hand this to a client that auto-follows redirects and discards the hops, and you get page 0 and no way to ask for page 1. ### Stateful pagination Page 0 is a fixed ten-job preview. More jobs come from `?next&rowFrom=`, but only **inside the session that served page 0**. It's a jScroll infinite scroll, and the session cookie is what makes it work. A cookieless `?next` returns nothing. Not an error, not a `401`: an empty page that looks exactly like the end of a board. A client that doesn't thread cookies through the pagination loop reads every TBE board as having exactly ten jobs, forever, at `200 OK`. Every board, plausibly populated, silently truncated. If your Taleo numbers look suspiciously round, count them. ## What else bites you **There are two themes and only one has JSON-LD.** The modern theme emits schema.org `JobPosting` on the detail page. The older one emits none at all, renders the body in a `
` block instead, and packs the location into the listing anchor as `Title | City`. Handle only the modern theme and a chunk of boards come back with titles and no descriptions. **A header block sits above the description.** On some tenants a job-description *header* element precedes the real body, and it's a ~150-char teaser. Take the first match and you store the teaser, which is populated, plausible, and wrong. **`rid` is unique only within an org.** Two organisations can both have the same requisition number, so the key has to be scoped by org or one silently overwrites the other. **The employer's domain is never in the JSON-LD.** TBE's `hiringOrganization` is name-only: no `sameAs`, no `url`. The domain does exist on the detail *page*, in its outbound links, mixed in with social and tracker hosts. Mining it at scrape time and corroborating it against the employer name is the only cheap chance you get at it. **A `403` here is a burst, not a wall.** TBE `302`s across shards as normal operation, and a burst can draw a `403` from the CDN. Both recover on the next attempt, so treating that `403` as permanent disables live boards. ## Or use ours The fetch is a one-liner. The session, the shard and the two themes aren't. You get descriptions on 710 of 712 rows, taken from JSON-LD where the theme has it and the `cwsJobDescription` block where it doesn't, with the teaser header kept from winning. You get the real employer name off the JSON-LD rather than an org code like `NORTHWIND2`, the domain mined from the detail page and corroborated against that name, and the [company join](/docs/objects/company) hanging off it. You get pagination that carries its cookie, so boards aren't truncated at ten. And Taleo rows are [collapsed to one row](/docs/guides/deduplication) against the same job on [LinkedIn](/ats/linkedin) rather than returned twice. Our queue seeds 700+ TBE boards, assembled rather than fetched, like most of this dataset. See [coverage](/docs/reference/coverage) for how often it's re-read and [sources](/docs/reference/sources) for the rest of the vocabulary. Query it with `?source=taleo`. [Quickstart](/docs/quickstart), then [pricing](/pricing).