🇮🇪 IrishJobs.ie Scraper — Jobs, Salaries & Employer Leads

Turn Ireland's national job board into a clean spreadsheet of hiring companies. Search IrishJobs.ie by keyword, county, contract type, recency, salary and advertiser type, and get back one structured row per live posting: job title, employer, employer ID, location, salary, the exact posting timestamp, employment type, industry, GPS coordinates and the full job description.

No login. No cookies. No account. Built on the site's own embedded result JSON, not brittle HTML scraping.

---

Why people use it

• 🎯 Recruitment & staffing agencies — a live "who is hiring right now" list for Ireland, filtered to direct employers only so you are not prospecting your own competitors.
• 📈 Sales & GTM teams — hiring is the loudest public buying signal there is. "Posted in the last 24 hours" plus a job title is a qualified trigger event.
• 🧠 Market & salary research — 65 distinct industry labels, structured pay bands, and exact posted/expiry timestamps across the whole board.
• 🤖 AI / RAG pipelines — full job descriptions as clean text or HTML, averaging ~7,200 characters per posting.

---

What each row contains

From the search results (100% populated unless noted, n=240 measured 2026-09-04):

• 🆔 job_id — the site's numeric job id, the stable dedupe key (100%)
• 🏷️ title (100%) · 🔗 job_url (100%) · ✂️ snippet (100%)
• 🏢 company (100%) · company_id — a stable employer id you can group on (100%) · company_profile_url (100%) · company_logo (100%)
• 📍 location (100%) · post_code (12% — mostly blank on this site, see Honest limits)
• 💶 salary_raw (100% present, but only 35% carry a number — see Honest limits) plus parsed salary_min / salary_max / salary_currency / salary_period (35%) and a salary_is_disclosed flag so you can filter cleanly
• 📅 date_posted — an exact ISO timestamp, not "3 days ago" (100%) · publish_from_date · publish_to_date (listing expiry)
• 🧭 section — main vs the site's padded semantic/regional cards (see Billing integrity)

From the job's own page (with "Fetch full job details" on — 240/240 parsed in the measured sample):

• 📝 description_text and/or description_html (100%, avg 7,171 chars of text)
• 💼 employment_type (100%) — FULL_TIME / CONTRACTOR / PART_TIME / TEMPORARY / APPRENTICE / REMOTE_WORKING
• 🏭 industry (99.6%) — e.g. "Accountancy, Accountancy-Accountancy"; 65 distinct labels in a 240-row sample
• ⏳ valid_through — the posting's expiry date (100%)
• 🗺️ latitude / longitude (100%) · address_locality (90%) · address_region (74%) · address_country (100%) · address_postal_code (12%)
• 🏠 is_remote + job_location_type — the site's real remote signal (9.6% of the sample were TELECOMMUTE)
• ✅ direct_apply (100%)

---

Quick start

{
  "searchTerms": ["accountant"],
  "sortBy": "date",
  "maxItems": 25,
  "scrapeJobDetails": true
}

The whole board, newest first: leave searchTerms empty — the actor sweeps /jobs/all (9,869 live postings measured 2026-09-04).

The weekly "who just started hiring" list for an agency BD team:

{
  "postedWithin": "7",
  "companyTypes": "directEmployer",
  "dedupeByEmployer": true,
  "fullCoverageMode": true,
  "maxItems": 0
}

A real hiring-trigger monitor — schedule it daily with "onlyNewSince": "lastRun" and it returns only postings that are genuinely new since the previous run (the watermark lives in a named key-value store, so it survives across runs).

---

Input reference

What to scrape

Field	What it does
searchTerms	Keywords, one per line. Empty = the whole live board (/jobs/all).
locations	Irish city, county or province — dublin, county-cork, munster, northern-ireland. Crossed with every search term.
startUrls	Paste any irishjobs.ie search URL straight from your browser with facets already applied — the actor uses its path and query as-is. A single /job/... URL is also accepted.

Filters (applied on the site, so you only pay for rows you asked for)

Field	What it does
jobType	permanent · work-from-home · contract · part-time · temporary
postedWithin	1 / 3 / 7 / 14 days. Board-wide today: 2,809 new in 24h, 6,087 in 3 days, 8,200 in 7 days, 9,620 in 14 days.
companyTypes	directEmployer (6,308 board-wide) or agency (3,561). excludeAgencies is a shorthand for the former.
minSalary + salaryPeriod	Minimum advertised pay, per year / day / hour.
sortBy	relevance · date · salaryHigh · salaryLow

Scope & coverage

Field	What it does
maxItems	Hard cap on rows returned and charged. 0 = as far as the crawl plan reaches.
maxPagesPerQuery	Pages of 25 per search. Hard-capped at 4 — see The 100-row wall.
fullCoverageMode	Automatically slices an oversized search using the site's own facet counts and unions the results.
includeRelatedJobs	Keep the padded semantic/regional cards. Off by default — see Billing integrity.

Output shaping

Field	What it does
scrapeJobDetails	Fetch each job's own page for description, industry, employment type, GPS, expiry, remote flag.
descriptionFormat	both · text · html
dedupeByEmployer	One row per hiring company. Measured: 240 jobs collapse to 143 employers (a 40% collapse).
onlyNewSince	An ISO date, or lastRun for a stateful "only what's new" monitor.
employerNameFilter / excludeEmployers	Case-insensitive company-name include / exclude.
titleMustInclude / titleMustExclude	Case-insensitive job-title include / exclude.
outputFields	Keep only these columns, in this order — narrow CSVs for a sales team.
flattenOutput	Flatten nested values for Excel / Google Sheets.
includeRawJson	Attach the untouched source records under raw.

Advanced

proxyConfiguration (defaults to Apify RESIDENTIAL, country IE — required, see below) · maxConcurrency · maxRequestRetries · diagnose

---

The 100-row wall, and how this actor gets past it honestly

IrishJobs hard-resets the connection on ?page=5 and beyond. This was measured on three different queries, from a home IP and from an Irish residential proxy, over HTTP/1.1 and HTTP/2, as the first request of a fresh process. One search therefore reaches 4 pages = 100 rows, and this actor clamps there rather than logging errors on every run.

The site's own "next page" link advertises ?of=25. That parameter is a decoy. It returns HTTP 200 with a full page of cards, and pagination.page stays 1 — a 20-request contiguous crawl of of=0…475 returned 500 rows of which only 32 were unique (93.6% duplicates). An actor that trusts the site's next-link ships duplicates and bills you for them. This actor verifies every page against pagination.page and discards anything the site did not actually paginate.

fullCoverageMode goes wider the legitimate way. It crawls all four sort windows (measured near-disjoint: sort=2 overlapped relevance page 1 by 1/25, sort=3 by 1/25, sort=4 by 0/25), then reads each facet's live count from the site's own filter contract and splits the query into slices small enough to fit the window — location, then recency, then job type, then salary band.

Measured on /jobs/accountant, a query the site reports as 442 results (423 of them main):

Mode	Rows returned	Requests	Wall time
Default	92	4	9s
fullCoverageMode: true	419 unique (99.1% of the 423 real results)	360	393s

---

Billing integrity — what this actor refuses to charge you for

1. Padded rows. IrishJobs injects semantic and regional cards into narrow searches. /jobs/chef/in-galway reports 20 results and renders 12 real + 8 padded. Those 8 are dropped and never billed unless you ask for them with includeRelatedJobs.
2. Duplicates. Sort-window coverage deliberately re-fetches the same jobs. Measured: 100 raw cards collapsed to 50 unique rows in a duplicate-input test, and 650 raw rows to 419 unique in the coverage run. Deduplication by job_id happens before the charge, not after.
3. Over-cap rows. Rows are delivered and billed atomically, so hitting your maxTotalChargeUsd cap can never leave you holding rows you were not charged for, or charged for rows you did not receive.

---

Pricing

Event	Price	When it fires
Per job returned	$0.0006	Once per deduplicated job saved to your dataset
Per job detail fetched	$0.0006	Only when "Fetch full job details" is on

A full-detail row costs $0.0012 — cheaper than every IrishJobs actor on the store today, and there is no per-run start charge. 1,000 fully-enriched postings = $1.20.

---

Honest limits

• Salary is mostly not disclosed. salary_raw is present on 100% of rows, but only 35% (83/240 measured) contain an actual number — the rest read "€ Not Disclosed". There is no structured salary anywhere on the wire: the JobPosting JSON-LD carries no baseSalary even on postings that display a full €95,000–€105,000 range, so salary_min/salary_max are parsed from the displayed string. salary_is_disclosed tells you which rows are real.
• post_code and address_postal_code are ~12% filled, and street_address is empty on 100% of the sample. Use latitude/longitude (100%) for geography.
• Fields we deliberately do NOT advertise. isSponsored, isHighlighted, isTopJob, isAnonymous, workFromHome, partnership.isBackfilled, isCrossPosted, skills[] and unifiedSalary all exist in the source schema and were 0% populated across 240 rows. They are passed through as-is rather than dressed up as features. The real remote signal is the detail page's jobLocationType: TELECOMMUTE (9.6% of the sample), surfaced as is_remote.
• A residential Irish IP is required. Measured: Apify's datacenter pool and shared auto pool both receive a 389-byte Akamai 403; Irish residential returns real data. The default proxyConfiguration handles this — if you override it, keep RESIDENTIAL.
• Reliability. Across the sustained validation runs: 359/360 SERP requests usable in the coverage run and 250/250 in the field-fill run. Transient NGHTTP2_INTERNAL_ERROR and proxy 5xx errors occur at roughly 5% and are cured by the built-in fresh-session retry; none reached the output.
• This is live-listing data. Postings expire — publish_to_date and valid_through tell you when. A run is a snapshot, not an archive.

---

Legal & fair use

This actor reads only public, unauthenticated IrishJobs.ie pages — the same pages any visitor sees. No login, no account, no paywall, no anti-abuse control is touched.

irishjobs.ie/robots.txt has no blanket Disallow: / in its User-agent: * group (the blanket disallows sit in the named-agent groups for adidxbot/Twitterbot and Applebot/Bing/Slurp). Quoted verbatim, the wildcard-group directives that touch the paths this actor uses are:

User-agent: *

Allow: /jobs/work-from-home
Allow: /jobs/temporary
Disallow: /jobs/permanent
Allow: /jobs/contract
Allow: /jobs/part-time

Disallow: /jobs/work-from-home/*/in-*
Disallow: /jobs/temporary/*/in-*
Disallow: /jobs/contract/*/in-*
Disallow: /jobs/part-time/*/in-*

Disallow: /*&page=  # this is where there is a page parameter on a facited link - its a catch all
Disallow: /*?action=
Disallow: /*&action=
Disallow: /*salary=
Disallow: /*jobtype=
Disallow: /*postedwithin=
Disallow: /*/apply?
Disallow: /*/apply$

Note that Disallow: /*&page= covers a paged URL that also carries a sort or facet parameter (e.g. ?sort=2&page=3), which this actor uses; /jobs/permanent and the /jobs/<type>/*/in-* combinations are likewise disallowed for the wildcard group. Deciding whether to run this actor is your call, and complying with irishjobs.ie's Terms of Service is your responsibility. The actor never touches /apply and strips the site's action= tracking parameter from every request it makes.

Job postings can contain personal data (named recruiters, contact details inside descriptions). Handling that data lawfully — GDPR included, since this is Irish and EU data — is the operator's responsibility.

---

FAQ

Do I need an IrishJobs account or login? No. Everything comes from public pages.

Why is it capped at 4 pages per search? Because the site resets the connection at page 5 — reproduced from three networks and both HTTP versions. Turn on fullCoverageMode to reach the rest of a large result set through facet slicing instead.

How do I get every job on the board? Leave searchTerms empty, set maxItems: 0 and fullCoverageMode: true. Expect a long run and raise the run timeout accordingly.

How do I get only new jobs each day? Set onlyNewSince: "lastRun" and add a daily schedule. The actor stores a per-search watermark in a named key-value store, so a second run minutes later returns 0 rows (verified).

Can I get one row per company instead of per job? Yes — dedupeByEmployer: true. Measured: 240 jobs → 143 employers.

Why do I sometimes get fewer rows than the site's result count? Because the site's count includes semantic/regional padding you did not search for. Those rows are dropped and not charged. Set includeRelatedJobs: true to keep them, tagged with their section.

Is the salary reliable? Only ~35% of postings state one. Filter on salary_is_disclosed rather than assuming a number exists.

Can I export to CSV / Google Sheets? Yes — the Dataset tab exports CSV, JSON, Excel and more. Use outputFields and flattenOutput for a narrow, spreadsheet-friendly file.

Does it work for Northern Ireland? Yes — locations: ["northern-ireland"] is a valid facet on the site.

---

Related actors

Same parser family, different national board: Reed.co.uk Jobs Scraper (UK), Totaljobs Scraper (UK), Jora Jobs Scraper (AU), JobStreet Jobs Scraper (SE Asia).