Product

Stop running your own job processing pipeline.

ClassifyCore ingests, cleans, classifies and enriches jobs at any scale — from 1,000 to 10 million postings. Delivered to your job board via XML or API.

Request a Quote Talk to Alexander

A full-service pipeline that ingests, cleans, classifies and enriches job data at scale. Built for job boards, aggregators, ATS platforms and programmatic vendors who need clean, structured jobs delivered to their site — without building or maintaining the infrastructure themselves.

The Problem

Running a job board or aggregator means moving job data from dozens or hundreds of sources into a single, clean, classified, searchable inventory. The operational cost is rarely budgeted properly:

  • Feeds arrive in inconsistent XML schemas, broken RSS, partial APIs, and bespoke integrations. Each source breaks differently and at a different cadence.
  • Duplicates compound across sources. The same role posted by an employer, two agencies and an aggregator becomes four listings, four indexed pages, and four hits on Google's helpful content radar.
  • Classification done on the job title alone misses the actual content of the role. Seniority, employment type, remote arrangement, salary, licences, education requirements — all sit inside the description and most platforms never extract them.
  • The engineering team that should be building product ends up maintaining scrapers, dedup rules, taxonomy mappings and feed monitoring.

The result is a thinner inventory than the raw feed count suggests, weaker Google Jobs schema, fewer filters on category pages, lower-quality email alerts, and an engineering team distracted from the things that actually grow revenue.

How It Works

ClassifyCore is a four-stage pipeline. You point your sources at it and tell it where to deliver. The infrastructure, models and operational monitoring sit on our side.

ClassifyCore pipeline: Ingest, Process, Enrich, Deliver.

1. Ingest — Any source: XML feeds, JSON APIs, RSS, ATS integrations, partner feeds, served APIs. New sources are onboarded without engineering effort on your side.

2. Process — Normalize, deduplicate, filter and prioritize. Duplicate logic operates across sources, not just within them, so the same role from an employer + two agencies + an aggregator collapses to one canonical posting.

3. Enrich — Classify against any official taxonomy (ESCO, O*NET, SOC, UKSSC, NOC, KLDB) or a custom client-specific taxonomy. Extract structured attributes from the full job description, not just the title.

4. Deliver — Unified XML feed, direct API push, or native integration with major job board SaaS platforms.

What ClassifyCore Actually Does

1. Any enrichment, from the full job description

Most classification tools read the title and stop. ClassifyCore reads the full description and extracts structured attributes that should live in your schema, your filters and your email alerts but usually don't:

  • Taxonomy mapping — ESCO, O*NET, SOC, UKSSC, NOC, KLDB, plus custom client-specific taxonomies for niche markets
  • Employment type — full-time, part-time, contract, temporary, internship, apprenticeship
  • Seniority — entry, junior, mid, senior, lead, executive
  • Salary and salary band — extracted from any format and normalized to your currency and period
  • Remote type — on-site, hybrid, fully remote
  • Extended remote — specific in-office vs. home-office day patterns where stated
  • Shift information — day, night, rotating, weekend
  • Licences and certifications — extracted and normalized
  • Education and degree requirements
  • Address enrichment — geocoding, region/county/metro mapping
  • Benefits — extracted and categorized
  • Contact Details — extracted and remove contact detals, phone numbers, emails

If it's in the job description, we extract it and normalize it. The downstream effect: complete Google JobPosting schema, category pages that target long-tail intent ("registered nurse night shift Manchester"), more filters on search pages, and email alerts that actually match what the candidate asked for.

2. No infrastructure required

Scrapers, feed monitoring, deduplication rules, classifier retraining, taxonomy maintenance, multilingual coverage — none of it lives on your side. You tell us where to deliver. We handle everything before that point.

Scales from 1,000 jobs to 10 million. Same pipeline, same SLAs. Your engineering team works on product, not on the gap between a vendor's XML feed and your search index.

3. Cleaner inventory than your raw feed

Extensive cross-source deduplication collapses expanded and duplicated jobs that inflate inventory counts without adding value. Custom filtering can be applied at any field — title patterns, locations, sectors, employment types, sources — so you only receive jobs that fit your niche. Multilingual: any language in, normalized output.

Use Cases

  • Job aggregators — Onboard hundreds of sources, deduplicate across them, classify into a unified taxonomy, deliver one clean feed to your search index.
  • Niche job boards — Filter a broad source down to your sector, enrich with the attributes your candidates actually filter on, and build category pages around real long-tail intent.
  • HCM and ATS platforms — Standardize job data flowing through your platform with consistent taxonomy mapping and enrichment.
  • Recruiting agencies — Process and classify incoming requisitions at scale, route by skill and location, surface gaps.
  • Programmatic job ad vendors — Clean, classify and enrich publisher inventory before bid logic runs.

Where It's Deployed

ClassifyCore's pipeline and underlying technology are in production across:

  • crypto-careers.com and web3jobs.io — niche aggregators in Web3 / crypto
  • jobsgopublic.com and lgjobs.com — UK public sector
  • nurses.co.uk, healthjobs.co.uk and healthcarejobs.ie — healthcare
  • prosple.com — early careers, multi-region
  • Multiple Jobiqo client deployments across Europe

Pricing and Engagement

ClassifyCore is priced per deployment based on volume, number of sources, taxonomy complexity and delivery format. Engagements typically include:

  • Scoping call to map sources, target taxonomy and delivery format
  • Onboarding of initial sources and taxonomy configuration
  • Custom classifier training where the use case warrants it
  • Ongoing pipeline operation, monitoring and source maintenance

Request a quote below with a short description of what you're trying to solve and we'll come back with scope and pricing.

Related Services

Frequently Asked Questions

How often can the data be updated?
As often as you need it, down to hourly. Most clients run on a daily cycle because most source feeds only refresh daily and there's no reason to spend cycles re-processing identical data. Aggregators with high-velocity sources — programmatic publishers, real-time ATS feeds — run hourly.
Can you handle non-English job feeds?
Yes, any language. Classification, deduplication and field extraction all run on the source language — we don't translate first and lose nuance. Multilingual clients typically run a mixed inventory (German, French, Spanish, English in one feed) and ClassifyCore handles each posting in its native language.
How long does onboarding take, and what's the work on our side?
A couple of days for a simple deployment and up to two weeks for complex deployments with many sources, a custom taxonomy, or a native integration into your platform. The variable is usually testing speed on your side, not ours. The work on your side is: handing over feed credentials or API access, agreeing the taxonomy and dedup policy, and reviewing the test output before we flip the switch on the live feed.
Can you extract custom information from jobs that isn't on your standard list?
Yes. If it's in the job description, we can extract it. Clients have asked for things like security clearance level, language requirements, specific software stacks, union membership, sponsorship eligibility, on-call patterns, even dress code. As long as employers actually write it into descriptions, we can build an extractor for it. The list on the product page is an example — anything beyond that is a configuration decision during onboarding.
How accurate is the classification, and can we audit it?
Accuracy depends on the field and the taxonomy. For coarse fields like employment type or seniority, we run in the high 90s on most clients. For occupation mapping into ESCO or O*NET, the realistic range is 85–90%, depending on the granularity of the target taxonomy. Custom client-specific taxonomies trained on your labeled examples land higher. You can audit two ways: every deployment gets an API endpoint that returns the classification plus a confidence score per field, so you can pull random samples and inspect them. We can also run blind tests against your data so you can evaluate the quality yourself.
Can you source organic jobs from employer career pages for us?
Yes. Sourcing organic jobs directly from employer career pages is part of the service. We need a few basic parameters from you — target employers or domain lists, the locations or roles you care about, the volume you want — and we'll scrape, normalize and run them through the same pipeline as your other sources. This is priced separately on top of the core pipeline.
Does this work with our existing job board software, and what happens if we want to leave?
On integration: ClassifyCore delivers to anything that can consume XML, JSON or a REST API, which covers every job board platform I've worked with — Jboard.io, Jobiqo, Jobboard.io, SmartJobBoards, Madgex, plus any custom platforms. On exit: you own the enriched data we deliver to you. There's no lock-in on the output side — your delivered feed is yours, your taxonomy mappings are yours, and we'll provide a final export if you offboard. The thing you don't get to keep is the pipeline itself (the source connectors, dedup state, trained classifiers), since that's the product.

Request a Quote

Tell us what you're trying to solve. Alexander will reply within 1–2 working days with scope and a quote.

Clean, classified job data, delivered.

Tell us what you're trying to solve. We'll come back with scope and a quote.

Request a Quote Book a Call with Alexander