Know a company with years of records? Refer it and earn up to $100K
Staffing and recruiting agencies

Staffing Agencies: License Candidate and Placement Records

Yes, with privacy handled first. How staffing agencies license Bullhorn job orders, submittals and placements to AI companies, and what buyers reject.

Short answerChecked

Yes, staffing agencies can license candidate, job order and placement records to AI companies, where deals for company records run $20K to $5M, once candidate privacy is handled. Buyers want the linked history: the job, who was submitted, the interview notes and who was placed. Names, contact details and raw resumes come out first, and California and EU candidates need a notice that covers the new use.

House keys on a lease folder and an inspection checklist
Key points
  • The value is the chain from job order to submittal to placement, not a pile of resumes.
  • California's privacy law covers job applicants, and the GDPR asks for a notice before candidate data is used for a new purpose.
  • Bullhorn exports through CSV list views, the REST API or a full SQL Server backup; plan the backup before any contract ends.
  • Strip names, contact details, resume files and client-confidential terms before anyone outside the agency sees a sample.

Which privacy rules govern candidate data at staffing agencies?

Candidate records are personal data before they are anything else. Start with the rules below, read on their official sites. They decide whether a record can leave the agency at all, and in what form.

Rule or sourceWhat it saysWhat it means for youDate checked
CCPA, Cal. Civ. Code 1798.140(i)A consumer "means a natural person who is a California resident"Every California candidate in the ATS is a consumer under the law, not only shoppers20 Nov 2026
CCPA, 1798.140(v)(1)(I)Personal information includes "Professional or employment-related information."Work history, pay, placements and interview notes are covered20 Nov 2026
California Privacy Protection Agency FAQRights cover "California residents that are employees or job applicants"The old employee exemption is no shelter for applicant files20 Nov 2026
CCPA, 1798.140(ad), "sell"Disclosing personal information "to a third party for monetary or other valuable consideration"Licensing identifiable candidate records for money is a sale20 Nov 2026
CCPA, 1798.140(m), "deidentified"Information that "cannot reasonably be used to infer information about, or otherwise be linked to, a particular consumer"Properly de-identified records fall outside the sale rules, with conditions20 Nov 2026
GDPR Article 14(4)Before further processing for another purpose, the controller provides "information on that other purpose"EU candidates sourced from job boards or referrals need a new notice first20 Nov 2026
GDPR Recital 26The Regulation "does not therefore concern the processing of such anonymous information"Truly anonymous data is outside the GDPR; pseudonymized data is not20 Nov 2026

Sources: Cal. Civ. Code 1798.140, the CPPA FAQ and the GDPR on EUR-Lex, read on 20 November 2026. This is general information, not legal advice. For the wider privacy picture, see GDPR, CCPA and licensing company data.

Can a staffing agency license its candidate and placement records to AI companies?

Yes, and the reason is the outcome data. A resume on its own says who a person claims to be. A placement record says that a recruiter judged a person fit for a specific job, that a client agreed, and that the person started. Thousands of those decisions, with the job description beside them and the rejected candidates too, are what a model needs to learn matching and screening. Public job boards and scraped profiles do not carry the outcome.

Demand comes from companies building hiring software, and the law now pushes them toward better data. The EU AI Act lists "AI systems intended to be used for the recruitment or selection of natural persons" as high-risk in Annex III, and Article 10(3) requires their training data to be "relevant, sufficiently representative, and to the best extent possible, free of errors and complete in view of the intended purpose." Article 10 also asks for an "examination in view of possible biases". In New York City, Local Law 144 already bars employers and employment agencies from using an automated employment decision tool without a bias audit in the past year.

Those rules make real recruiting histories, across many roles, industries and years, more useful to a developer than synthetic or scraped data. They also make the developer careful. A buyer will ask how candidates were told, how records were de-identified and whether protected characteristics were collected at all. Sources: the AI Act and NYC DCWP.

What is in a Bullhorn job order, submittal and placement record?

Bullhorn's own API reference describes the chain. A JobOrder "Represents an open job to be filled." A Sendout "occurs when a Candidate's information is sent to a ClientCorporation to be evaluated for a particular job." A JobSubmission "Represents a formal submission of a Candidate for a particular job", created after the candidate has been assessed. A Placement "Represents a successfully filled job", and each one "corresponds to a single JobOrder and a single Candidate." A Note can hang off a Candidate, a JobOrder and other records, which is where interview notes and call summaries usually live.

That linkage is the dataset. Pull a job order and you can follow every candidate sent, every submittal, the notes on each, and the one or two people placed. Agencies on JobAdder, Vincere, Crelate or a Salesforce-based ATS hold the same shape under different names: a vacancy, an application or shortlist, activity notes and a placement or start record.

Check three things in your own system before describing it. First, how far back the job orders go with their status history intact. Second, whether submittals and placements still point to the right job order after old migrations. Third, how much of the interview record is in structured fields and how much sits in free-text notes or attached files. Source: the Bullhorn entity reference.

What do AI companies want from staffing records, and what do they reject?

They want decisions with context. The useful unit is a job order with its description, requirements, client industry and location band, linked to every candidate considered, the stage each one reached, and the reason they stopped. Interview notes and recruiter feedback add the most, because they record why. Volume matters: years of job orders across many clients beat a large but shallow resume store.

  • Wanted: job descriptions and requirements tied to their outcomes (filled, lost, cancelled), with dates.
  • Wanted: submittal and interview stages per candidate, with recruiter and client feedback in text.
  • Wanted: placement records with start dates, role type and whether the placement lasted, where the ATS tracks it.
  • Wanted: skills, titles and experience in structured fields, after names and contact details are removed.
  • Rejected: raw resume files with names, emails, phone numbers and addresses still inside.
  • Rejected: records of candidates who were never given a notice covering this use, where the law requires one.
  • Rejected: client rates, margins and contract terms the agency agreed to keep confidential.
  • Rejected: fields that reveal protected characteristics, such as age, health or nationality, unless the buyer asked for them for bias testing under a lawful basis.

Resume text is the hardest part to clean. Names, schools, small employers and towns can identify a person even with the header removed. Many agencies license structured fields and notes first, and treat resume text as a separate, later decision.

How do you export data from Bullhorn and other ATS systems?

Bullhorn offers three routes, and its knowledge base explains when each fits. For "a few thousand records", exporting a list view to CSV "is the most efficient method." Export limits per entity type range "from 5,000 to 20,000" records at once, depending on the user's entitlement. For more, "Access to the Bullhorn API is available to Front Office Growth or Enterprise editions of Bullhorn." For everything, an account contact can request a data backup.

The backup is the route most agencies use for a full history. Bullhorn says it "consists of a Microsoft SQL Server generated file containing all of your company's data". A final or full backup "Includes all data, such as records and notes, plus all file attachments in their native format, such as resumes in Microsoft Word format." Bullhorn also warns that "we may not retain your data past 60 days of your contract ending", so an agency closing down or switching ATS should request the final backup before that window closes.

  • Decide the scope: which years, which entities (job orders, sendouts, submittals, placements, notes) and which regions.
  • For a sample, export list views to CSV from each entity and keep the record ids so the links survive.
  • For the full history, request a backup through your Bullhorn account contacts and restore it to a database you control.
  • Before anything leaves, drop names, emails, phones, addresses, social links and resume attachments, and replace candidate and client ids with new random keys.
  • Scan note text for names, phone numbers and client names, and remove or mask them.
  • Keep the mapping from new keys to old ids inside the agency, never in the licensed files.

Sources: Bullhorn's export guide and data backup guide. For what a resume database holds and a worked export, see what a staffing agency ATS really holds; agencies on a Salesforce-based ATS can follow exporting Salesforce data.

Should staffing agencies license records before they sell or close?

Often the timing decides it. An owner selling a staffing or recruitment agency is usually thinking about the business: clients, recruiters, revenue and a staffing company valuation. A data license is separate. It can be signed before a sale, by a closing agency, or kept out of the deal entirely, but it has to fit the sale terms if a buyer of the agency expects the database to come with the business.

Closing raises the stakes on both ends. The ATS contract ends and, in Bullhorn's case, the data may be gone 60 days later. Privacy law pushes the other way: records kept past their purpose are a liability, not an asset, especially for EU candidates. An agency that is closing should take the final backup, decide quickly what it may lawfully keep and license in de-identified form, and delete the rest on a documented schedule.

An agency being sold should tell the acquirer about any license, because the acquirer may want the same records for its own systems or may object to a non-exclusive license running alongside its ownership.

What should a staffing agency do next?

We help you license your data. Origin Data Partners is an independent data partner that helps business owners value and license their company records: the company pays Origin nothing, and a buyer may pay Origin when a deal completes.

Start with a description, not files: which ATS, how many years of job orders, rough counts of submittals and placements, how much interview text sits in notes, which countries your candidates live in, and what your candidate privacy notice says. No candidate record needs to leave the agency for that first look.

We work only with data you created in running your business, or hold under a written licence that lets you license it for AI training. We do not take scraped web data, purchased contact or consumer lists, people-search or location data, or personal data collected without a notice that covers this use.

When you have the outline, see what your records may be worth. Before signing anything, read what to check in a data licensing offer and what goes in a data licensing agreement. Not the owner? Advisers can introduce the company.

How we checked this

We read every source on this page on 20 November 2026. The rules come from primary law on official sites: the California Civil Code on the state legislature's site, the California Privacy Protection Agency's own FAQ, and the GDPR and the EU AI Act on EUR-Lex. The Bullhorn facts come from the company's own API reference and knowledge base. The New York City rule comes from the city regulator's page.

What we could not verify: any price paid for staffing or recruiting records, because no buyer or agency has published one we could find; the privacy notices of individual agencies, which differ; and how other ATS vendors package full backups, which their public help pages do not describe in the same detail as Bullhorn.

Common questions

Can a recruiting agency sell or license its candidate and placement data to AI companies? Who buys it and what does it pay?

Yes, in de-identified form or with the notices privacy law requires. Buyers are companies building hiring, matching and screening software, which need real decisions with outcomes. No buyer has published a price for recruiting records. Across company records, AI companies pay $20K to $5M, and the figure depends on years covered, how well job orders link to placements, and how much interview text survives cleaning.

We are a staffing agency with 15 years of resumes, job orders and placements. Do AI companies buy that?

The job orders and placements are the strong part. Fifteen years of vacancies linked to the candidates submitted and the person hired is rare outcome data. The resumes are weaker on their own and the hardest to clean, since names and employers identify people. Expect a buyer to ask for the linked records first, de-identified, and to treat resume text as an optional extra.

Who helps a recruiting or staffing agency license its data to AI companies?

Origin Data Partners helps staffing and recruiting agencies value their ATS records, prepare a de-identified description and license it on written terms. The agency pays Origin nothing; a buyer may pay Origin when a deal completes. Your privacy counsel should review candidate notices, and your ATS vendor handles the backup or API access. Start by describing the records rather than sending any.

What is a recruiting firm's resume database worth when we sell the business or close it?

As a licensed dataset, the value comes from linked outcomes, not resume count. Company records license for $20K to $5M, and a recruiting history sits higher in that range when job orders, submittals and placements link cleanly over many years. In a sale of the agency, the database is valued with the client book. When closing, act before the ATS contract ends.

Do we need candidate consent before licensing resumes and interview notes to AI companies?

For identifiable records, often yes, or at least a notice. California treats a paid disclosure as a sale and gives candidates rights over it. The GDPR requires information to EU candidates before their data is used for a new purpose. De-identified records avoid most of this if they truly cannot be linked back. Interview notes need the most cleaning, because they name people.

How do I export data from Bullhorn to Excel?

Open the list view for the record type you want, such as candidates, jobs or placements, filter it, select the records and choose the export to CSV. Bullhorn says CSV suits a few thousand records, with limits per entity type set by your entitlement. Open the CSV in Excel. For a full history with links between records, request a data backup instead.

Where is Bullhorn data stored?

In Bullhorn's hosted system, not on your own servers, which is why you reach it through the application, the REST API or a backup. A backup arrives as a Microsoft SQL Server file with all records and notes; a final backup adds attachments such as resumes in their native format. Bullhorn may not keep your data more than 60 days after a contract ends.

Is it legal for companies to sell your data?

Sometimes, under conditions. In California, selling personal information triggers notice and opt-out rights, and job applicants are covered. In the EU, a new use needs a lawful basis and information to the person first. From the agency side, the safer route is licensing de-identified records under a contract that bars re-identification. This is general information, not legal advice.

Sources and further reading

What is your company’s data worth?

AI companies pay $20K to $5M for company records. Six questions, no files.