Answers to the practical questions

A RoleDemand run has eight fixed stages: collect from the employer's applicant tracking endpoint, track the posting against yesterday's state, clean it into a canonical shape, classify function and seniority, resolve the location, scope the result to your grant, run quality checks, then rebuild and publish the dashboard.

1. Collect

   One adapter per applicant tracking platform calls the employer's own public endpoint. Network policy, retries and per-domain politeness live in one place, so no adapter can decide to be impolite on its own.

2. Track

   Each posting is matched against a lifecycle database, which turns a list of open roles into the two facts that matter: what opened since yesterday, and what closed.

3. Clean

   Titles, departments and locations are normalised into one canonical frame. A run whose columns do not match the schema fails rather than shipping.

4. Classify

   Function and seniority are resolved by a fixed cascade: a language model, then a trained classifier, then hand-written rules. Each posting records which of the three decided it.

5. Resolve locations

   Location text is resolved against a shared gazetteer, so one employer writing "London / Hybrid" and another writing "Greater London" land in the same place.

6. Scope

   Your grant is expanded into an explicit list of modules, sectors and employers. Anything outside it is dropped before the dashboard is built.

7. Check

   A quality report covers classification coverage, collection status per employer, circuit-breaker cooldowns, sudden drops to zero and company-name hygiene. Threshold breaches warn; schema mismatches stop the run.

8. Build and deliver

   The bundle is rebuilt for your scope and published behind a signed cookie. The briefing is written from the same cycle's numbers.

RoleDemand has 34 applicant tracking adapters, covering Greenhouse, Lever, Workday, Ashby, Teamtailor, SmartRecruiters, Personio, Workable, iCIMS, SAP SuccessFactors and 24 more. All 34 were probed against live public tenants on 6 August 2026 and all 34 returned a valid response.

• Ashbyashby
• BambooHRbamboohr
• Breezy HRbreezy
• CareerPuckcareerpuck
• JSON-LD behind a bot checkcf_jsonld
• Comeetcomeet
• Eightfold AIeightfold
• Gemgem
• Greenhousegreenhouse
• HiBobhibob
• Homerunhomerun
• iCIMSicims
• JazzHRjazzhr
• Jobvitejobvite
• JOIN.comjoin
• Generic JSON-LDjsonld
• Leverlever
• Manatalmanatal
• Oracle Cloud Recruitingoracle_orc
• OrangeHRMorangehrm
• PeopleForcepeopleforce
• Personiopersonio
• Phenom Peoplephenom
• Pinpointpinpoint
• Recruiteerecruitee
• Ripplingrippling
• SmartRecruiterssmartrecruiters
• SAP SuccessFactorssuccessfactors
• Oracle Taleo Business Editiontaleo
• Talexiotalexio
• Teamtailorteamtailor
• Workableworkable
• Workdayworkday
• Zoho Recruitzoho_recruit

An employer on a platform outside this list is still reachable. The two generic readers handle any careers page that publishes structured job data of its own, and adding a named platform means adding one file to the pipeline.

The run keeps going. A per-employer circuit breaker takes a failing source out of the rotation instead of letting it stall the cycle, the quality report names the employer and the failure, and repairing the source is part of what the monthly fee covers.

Sources move. A company migrates from one applicant tracking system to another and the endpoint that worked yesterday returns nothing today. That is the ordinary failure in this business, which is why repairing it is written into the price rather than billed as a change request. See what each plan covers.

Nothing. You name the market and the employers once, at the start. After that the run is automated, every cycle writes its own quality report, and repairing a source that moves is part of the service.

Your time goes into the setup conversation: which employers belong in the market, and how roles group for your desk. Setup is a conversation rather than a project, and it sits inside the monthly fee along with collection, quality checks and repairs.

Every posting comes from an employer's own public careers page or from the applicant tracking system endpoint that page is built on. Those are pages the employer published in order to be read, and RoleDemand reads them the way a person with a browser would, on a schedule and politely.

RoleDemand collects job postings, not people. There is no candidate database, no CV store, no contact scraping and no profile matching anywhere in the pipeline, and nothing behind a login is ever fetched. One qualification worth stating plainly: employers sometimes write a recruiter's email address or a direct phone number into the body of an advert. RoleDemand strips those as the posting is read, before the row is written, so they are not in the database, the export or the dashboard.

Never collected

• Candidate names, contact details, CVs or applications
• Anything behind a login, a paywall or an authenticated API
• Data from a job board that aggregates other people's postings
• Salary figures invented to fill a gap in a posting

It is a working RoleDemand dashboard that anyone can open without an account. It was built from the seeded renewable energy market on the 7 August 2026 cycle: 21 real employers whose public careers endpoints were checked by hand, read through the same eight pipeline stages a client cycle runs on. It is one file with the data inside it, so it opens in a single request and keeps working if the connection drops.

Renewable energy is the market published because its whole roster is public employers checked by hand, because nothing it advertises breaks the rule about what may appear on an open page, and because it is one of the widest packs to read: the pack spreads its 56 employers across 13 applicant tracking platforms today, none covering more than eight. Construction is the widest of the 11 now, at 15. Either is a market where reading every careers page by hand stops being tedious and starts being impossible. The published file is older than those numbers: it is frozen at the 7 August 2026 cycle, over the 21 employers the pack carried then. The renewable energy market page says what it covers now, and the dashboard itself is one click.

Start on This week, which opens on the roles that appeared and disappeared since the previous cycle. Then read the Employers ranking for the name you have never pitched, filter Demand to the role family you place into, and check Data health before quoting anything.

1. Start with what changed

   This week opens on the roles that appeared and disappeared since the previous cycle. That list is the call sheet.

2. Then look down the employer column

   Employers ranks tracked employers by how much they are advertising. A name near the top you have never pitched is the gap in your desk, stated as a number.

3. Then pick one role family

   Demand groups every posting by function and seniority. Filter to the family you place into and a market of hundreds becomes the twenty that are yours.

4. Check Data health before you quote anything

   It names any employer whose source has been failing, so you find out a number is thin before a client does. Too few rows is declared insufficient rather than smoothed.

Three things, with the published sample's 7 August 2026 cycle as the worked example. Who is moving: 9 of 21 tracked employers opened 3 or more roles in one discipline inside 30 days. What is hard to fill: 361 of 744 live roles had been advertised longer than 60 days. Where demand is going: 240 roles opened across 10 disciplines in 30 days.

It had been collecting for one day, so it declines to compare this month with last and says so where the comparison would have gone. A tracker reporting growth on one day of history would be inventing it.

Four things. It carries your agency's name rather than RoleDemand's, it tracks the employers you named rather than these ones, it is rebuilt every day rather than frozen, and it arrives with a written briefing each cycle. It is also scoped to your grant when it is built, so an employer outside your list has no row in the file your browser downloads.

The briefing is the part the file cannot show you: it is written from one cycle's numbers for one desk, and it is the short document your consultants forward to a client without editing. The 30-day cancellation window is measured against it.

Because a form in front of a sample is a way of collecting an email address from somebody who has not yet decided they want to talk. If the dashboard is any good it will do the persuading, and if it is not then an address taken before it loaded is worth nothing to either side.

Yes, and most clients will be one. The 11 seeded markets are worked examples rather than a catalogue. Adding a market means scoping a new one: naming the employers, checking their career pages, and agreeing how roles should be grouped for your desk. The pipeline does not change.

Every decision that makes a market yours: the employer roster, the geography it covers, the wording the dashboard uses, the way roles are grouped for your desk, and an optional lens that splits the market into two cohorts to compare. The core platform holds no market knowledge at all; it is all set when your market is scoped.

Keeping every market-specific decision in the market's own scope rather than in the pipeline is what makes a new market a scoping job. It is also what makes the entitlement rules enforceable: the platform cannot leak a market it does not know about.

The owner or director of a two to fifty person specialist recruitment agency who is also the business development function, and who has nobody in-house whose job is to analyse a market.

They know their market by name and by reputation, cannot afford an analyst, and are asked in client meetings to describe a market they have no systematic view of. Enterprise workforce-intelligence platforms do that job and are priced for corporate talent teams.

UK vacancies fell to 712,000 in April to June 2026, down 2.5 per cent on the year and down in 11 of 18 industry sectors, according to the Office for National Statistics on 21 July 2026. Fewer briefs are being placed, so more of the conversations you do get are competitive ones where a client is comparing you against another agency in the same week.

712,000

UK vacancies in April to June 2026, down 2.5 per cent on the yearSource: Office for National Statistics, 21 July 2026

11 of 18

industry sectors where vacancies fell over the yearSource: Office for National Statistics, 21 July 2026

£40.6bn

the recruitment industry's contribution to the UK economy in 2024, 1.6 per cent of gross value addedSource: REC

Scoping happens when the dashboard is built, not when it is requested. Your grant is expanded into an explicit list of employers, sectors and modules, and everything outside it is removed before the bundle exists, so an employer you are not entitled to see has no row in the file your browser downloads and there is no filtered view to defeat.

The same rule applies to modules: a section your agency has not bought is absent from the build rather than hidden in the interface. A grant is an explicit list rather than a wildcard, so adding an employer to the platform later cannot silently widen an existing client's access.

Yes. The subscription is per agency, not per seat, and each consultant signs in with their own email address. An individual account can be narrowed below the agency's grant, never widened above it.

A managed service, not software you operate. You choose the market and the employers that matter to your desk, and the dashboard and the briefing are built from exactly that roster. RoleDemand runs the daily collection, repairs sources when they move, checks the output, and delivers both. The work you do is naming the market once.

Setting a market up is the part that cannot be automated: agreeing the employer list, checking every careers page by hand, deciding how roles should be grouped for your desk, and putting your name on the dashboard. It sits inside the monthly price, because asking for money before you have seen a briefing is asking you to buy on trust you have no reason to give yet.

Stopping is an email. Cancel inside the first 30 days, after you have read the first briefing, and nothing further is charged. Beyond that window the minimum term is 3 months, the £169 founding rate is held for 12 months either way, and the dashboard stays readable until the end of the month you have paid for.

The window runs from the first briefing rather than from the day you signed anything: a product tour would show you the interface, and the first briefing shows you your own market, which is the fairer thing to judge.