Last checked: 7 October 2026. For companies, not individuals.
A recruiting firm holds two different things. One is a record of how searches get run: intake, sourcing, screening, client feedback, offers. The other is personal data about candidates. Buyers describe wanting the first. The second is where most of the risk sits.
AI systems are being trained and tested on how real work is done. Recruiting is almost entirely judgment, written down in many small steps.
micro1 says on its data partnerships page that “Operational data from every industry can contribute.” It lists SOPs, knowledge bases, internal documentation, CRM data, project histories and QA processes as the kinds of material it looks for, and it names “decision-making patterns” as something it values (as published, checked 7 October 2026). Grepped, which also pays individual professionals for their expertise, lists recruiters among the roles it is looking for, with “Sourcing, screens, offers” as the example work.
Put those together and the useful part of a staffing firm becomes clear. It is not the people in your database. It is the trail that shows how your team decides. Why was a requirement rewritten after the first client call? Why did the shortlist change after two screens? What did the account manager tell the client when a candidate declined? A single search can hold dozens of those decisions, linked by dates and stages.
That linked trail matters. Practitioners say raw data is the cheapest tier, and that evaluations built on top of the data are worth roughly ten times more. A tidy, connected history of searches is the kind of material that can be turned into tests of whether an AI system makes a sensible call. A folder of loose resumes is not.
Job intake notes, role scorecards, the questions you ask a hiring manager, and how a vague request became a clear brief.
Sourcing plans, search strings, channel choices and the reasoning when a plan changed midway.
Screening rubrics, structured debrief templates and the logic for advancing or holding a candidate, with identities removed.
How feedback from a client reshaped the search. Account management playbooks and escalation steps.
Offer process SOPs, approval steps and how your team handles counteroffers, written as process rather than as a named person’s file.
Onboarding checklists, compliance procedures, QA reviews of desk performance and internal training material.
The line is simple to state and hard to apply: your process can be in scope, the people in your pipeline should not be.
Each program publishes its own minimum. The question for staffing firms is which headcount counts: your internal team, or everyone on your payroll.
| Program | Published company payout | Published eligibility | Note for staffing firms |
|---|---|---|---|
| micro1 Enterprise Data Partnership | “$100k+ qualified”, “$500k+ large-scale”, “$1M+ highly unique” | 30+ employees, mature operations, documented processes, modern software tools, primarily English; U.S. prioritized, then other Western markets | The page does not say how placed workers are counted. Ask. |
| Mode company data | “$100K to $5M” | 20+ full-time U.S. office employees; several years of records the company owns; U.S.-based teams strongest fit | The minimum names office employees. Its calculator asks separately for desk and non-desk staff. |
| Grepped | “$20K to $5M” | Any vertical; also pays individual professionals for expertise | Lists recruiters (“Sourcing, screens, offers”) among the roles it seeks. |
Last checked 7 October 2026. Sources: each program’s own website (micro1 data partnerships and company referral pages, data.mode.inc, grepped.ai). Payout figures are published ranges, not offers. Miro Advisory also publishes indicative ranges for operating datasets ($100K to $1M+); it is not covered by the buttons below.
Independent site. Some links are referral links: if your company signs with a buyer through them, the buyer may pay us a fee. You are not charged, and we never see your data.
A 25-person agency may meet Mode’s published floor and fall short of micro1’s. A firm with 15 recruiters and 400 contractors on assignment is a harder case, because the published rules talk about employees and office staff, not placements. Do not assume. Run your numbers through the eligibility checker, then put the headcount question to the buyer in writing. The buyer programs comparison shows the full published terms side by side.
Staffing firms collect personal data for one purpose: placing people in jobs. Selling it, or a copy of it, for AI training is a different purpose. Several laws may apply, depending on where your candidates live and what you collected.
General information, not legal advice. Talk to your own lawyer before you sign.
This shows how to draw the line on paper before anyone exports anything. It has no price, because nobody can price data without reviewing it.
The published rules land differently on a boutique, a high-volume branch network and a multi-office group. These walkthroughs show how, without guessing at prices.
An executive search boutique. With 12 staff it is below Mode’s published 20+ office-employee minimum and micro1’s 30+. Grepped publishes no company size line and also pays individual professionals, so a senior partner’s own expertise may be a separate route. Its archive is also the most sensitive kind: confidential board-level searches. A realistic scope is narrow, such as its search methodology and interview frameworks.
A light-industrial agency that fills about 1,100 shifts a week. Its 28 internal staff clear Mode’s published floor on their own. They fall just short of micro1’s 30+ unless the buyer counts differently, which the firm asks about in writing. Useful records include order intake, fill-rate escalations and no-show handling. Associate payroll and work authorization files are excluded entirely.
Healthcare and IT staffing across 14 branches, so it clears both published minimums. More branches mean more systems and more exposure. Clinician credential files and health screenings stay out. Branch operations manuals, credentialing workflow SOPs and a pseudonymized client pipeline may stay in. A bigger archive is not automatically a bigger check: buyers price only after review.
Before talking to any buyer, list what exists. Most firms find that the includable part is smaller than the archive, and better organized than they expected.
| Record type | Typical system | Usually includable? | Why |
|---|---|---|---|
| Desk manuals and SOPs | Google Drive, SharePoint, Notion, Confluence | Usually yes | Written by you, about process, with little personal data. |
| Job orders and intake notes | ATS, email | Often, after scrubbing | Client names and confidential details must go; some clients forbid any reuse. |
| Submittal and placement stage history | ATS | Possibly, with pseudonyms | Shows the workflow clearly; every identity has to be replaced. |
| Interview debriefs and scorecards | ATS, Slack, email | Templates yes; filled-in notes rarely | Filled-in notes are opinions about named people. |
| Client sales pipeline | HubSpot, Salesforce | Possibly | Your own commercial process, with client contacts pseudonymized. |
| Internal team channels | Slack, Microsoft Teams | Process channels after review | Direct messages and candidate chatter stay out. |
| Resumes and candidate profiles | ATS, email attachments | No | Collected to place people, not to train AI systems. |
| Background checks, drug tests, I-9 files | Vendor portals, HR files | No | FCRA and employment records with strict handling rules. |
| Payroll and timesheets for placed workers | Paychex or another payroll system | No | Pay data tied to identifiable people. |
| Recorded screening calls | Zoom | No, in most cases | Recording consent and voice data. See call recordings and transcripts. |
If any answer below is “we don’t know,” settle it first. It is cheaper to settle now than halfway through a buyer review.
Owners often lead with candidate counts. Buyers describe wanting process and decisions, and the database is where the legal risk sits.
Claiming hundreds of placed workers toward an employee minimum can unravel late in review. Ask how headcount is counted, and keep the answer.
“Let the buyer filter it” sends personal data out of your control. Scope first, then export only the agreed slice.
Resumes ride along on email threads and tickets. A scope that excludes resumes must also exclude their attachments.
Recruiters type salary figures, health details and family situations into notes. Structured fields are easy to strip; free text needs review.
An exclusivity clause covering “all recruiting data” can block a later sale of a different slice. Match exclusivity to what you actually sold.
These apply to any buyer. They are not claims about any named program. Use them on whatever agreement you are offered.
Not as a list of people. Resumes, contact details, salary history and background checks are personal data about people who never agreed to this use. What buyers describe wanting is how the work gets done: SOPs, playbooks, project histories and decision-making patterns. A candidate database is the part to keep out, not the product.
The programs publish their own wording. Mode lists 20+ full-time U.S. office employees and micro1 lists 30+ employees (as published, checked 7 October 2026). Neither page we checked says how placed or temporary workers are counted, so ask the buyer before you rely on that headcount.
Connected histories that show judgment: an intake note, the sourcing plan, the screening rubric, the shortlist reasoning, client feedback and the final outcome for the same search. Scattered resumes or a single template are worth far less than a linked record of how a search moved from request to placement.
Commonly named ones include the Fair Credit Reporting Act for background check reports, state privacy laws such as CCPA/CPRA where applicants live in California, GDPR for any EU or UK candidates, and biometric privacy laws where video interview tools were used. Your client contracts and your own candidate privacy notice matter too. This is general information, not legal advice.
Probably not through the two company programs with published size lines. Mode lists 20+ full-time U.S. office employees and micro1 lists 30+ employees (as published, checked 7 October 2026). Grepped publishes no company size line and also pays individual professionals for their expertise, which may suit a senior recruiter better than a company deal.
Telling them before the scope is signed is the safer course. Their notes and messages are the core of what a buyer reviews, and they will notice an export. Say which systems and dates are in scope, what is excluded and why. Whether a notice is legally required depends on where they work, so ask your lawyer.
Plan for months, not weeks. Mode publishes that it generally expects about three months from the first conversation through payment, and practitioners cite 60 to 90 days to close. Scoping candidate data out carefully can add time, which is better than rushing it.
The checker runs in your browser and stores nothing. It shows each program’s published rule next to your numbers, including the employee minimums that matter most for staffing firms.