Read this first
A new kind of buyer has appeared in 2026: AI companies that want to license records of how established businesses actually do their work. Some of the headlines are large and some of the marketing is louder still. This guide sets out what the programmes themselves publish, what that means for companies, and the questions to settle before you share anything.
Every figure here comes from the programmes’ own websites, their own press releases, or (for micro1’s funding) the press report of its announcement. We checked each one on 30 October 2026 and list the sources at the end. Where a programme does not publish something, we say so rather than guess.
We may receive a referral fee from programmes.
We are independent. TrackImpact is not affiliated with, employed by or acting for any programme named in this guide, and none of them has reviewed or endorsed it.
We never see or handle your data. Programmes contract with you directly. Data Partnerships is separate from the TrackImpact AI-adoption platform, and platform customer data is never part of it.
This guide is general information. It isn’t legal, tax, financial, investment or other professional advice, and it isn’t a recommendation to license data or to join any programme. Figures are as each programme published them on the date shown; they can change, they aren’t offers, and nothing is guaranteed. Programmes decide who they accept, what they pay and on what terms. Check customer contracts, NDAs and data-protection obligations with your own advisers, and confirm everything with the programme, before sharing anything. To the extent the law allows, TrackImpact accepts no responsibility for decisions made on the basis of this guide, or for any programme’s terms, conduct or payments.
Can companies license their data to AI?
Yes, many established companies can, as of October 2026. AI companies are paying businesses to license internal operating records, and the material they can most easily take is process documentation: SOPs, playbooks, wikis, project histories and QA checklists. Three programmes publish terms aimed at ordinary companies. micro1 has committed $1bn over 12 months and publishes tiers from $100K–500K to $3M+. Mode publishes a range of $100K–$5M. Mercor publishes a range for each business type, from $10K to $1.5M. The usual entry point is 30+ staff with mainly English-language records; Mode accepts law firms from 6 staff and accounting, finance and medical firms from 10, and Mercor lists some business types from 5.
That paragraph is the whole story in brief. The rest of this guide covers the details that decide whether it is a good idea for your firm: what you would actually be licensing, what the programmes publish about value, who qualifies, the legal and professional checks, and the questions to ask before you sign.
What this is not
- Not selling your company or your clients. Programmes license an agreed copy of selected records for AI training. micro1 states that companies keep ownership; Mode buys an agreed, filtered copy and you keep your originals.
- Not a guaranteed payment. Programmes publish ranges. What any company is offered depends on its records, the programme’s current demand and the final contract.
- Not quick money with no work. Mode describes about three months from first conversation to payment, and every programme runs its own review before an offer.
Who is paying for company data, and why now?
AI companies are building agents that do real office work. Public web text shows what work looks like from the outside; agents also need records of how it gets done inside a business, step by step, with the exceptions and judgement calls left in.
That is why the programmes focus on established companies with years of documented process. Below is what each one publishes about itself.
micro1
- Committed $1bn over 12 months to license company operational data, with capital from Citi and Hercules Capital (announced 9 October 2026).
- Publishes three tiers: $100K–500K for a qualified partnership, $500K–3M for large-scale or multi-team data, and $3M+ for highly unique data.
- Looks for 30+ full-time employees with mainly English-language records. It starts with the US, then other Western markets; demand is strongest in the US, UK and Canada.
- States that companies keep ownership of their underlying data. micro1 de-identifies the data itself, so you don’t need to anonymise it first.
- Reviews and approves the data within 30 days of transfer, then pays within 30 days of invoice.
- Runs a separate Defunct Data Partnership for companies that are no longer operating.
Mode (mode.inc)
- Publishes a range of $100K–$5M per company. Its calculator can add a further amount for a data refresh over 24 months.
- Minimum size: 6+ staff for law firms, 10+ for accounting, finance and medical, and 30+ for everyone else. Teams based mainly in the US are the strongest fit.
- You keep your originals. Mode buys an agreed, filtered copy and resells it for AI research and training; your agreement sets the scope and any exclusivity. Your team signs in to the approved systems, and no passwords are handed over.
- Charges the company no broker fee, and describes about three months from first conversation to payment.
- Not to be confused with Mode Analytics (mode.com), an unrelated software company.
Mercor
- Publishes a range for each of 25 business types, from $20K–$250K (call centres) to $100K–$1.5M (manufacturers). Law firms are listed at $50K–$1.25M and accounting firms at $25K–$865K.
- Minimum size varies by business type, from 5 staff (law, accounting and investment firms) to 50 (construction).
- You connect workplace tools to Mercor’s platform. Nothing moves until you approve the scope; Mercor extracts and anonymises the data and pays by wire when the dataset is delivered, “typically within 2-4 weeks of the signed agreement”.
Data marketplaces: Protege and Defined.ai
Two other names come up in searches. They work differently: both run marketplaces that license data to AI developers and share the revenue with the data owner, and neither publishes a programme for company process documents in the way micro1, Mode and Mercor do.
- Protege (founded 2024; $65M raised, including a $30M round led by a16z) lists data-provider programmes for healthcare, video, audio and speech, spatial and physical data, and “other domains”. It pays a revenue share when your data is licensed, under agreed partnership terms; it does not publish a percentage.
- Defined.ai (founded 2015, Seattle) accepts audio, video, image, text and multimodal data with “no fixed minimum”, and pays licensing fees from marketplace sales. It currently highlights demand for cinematic video, anime, code and STEM data.
If your company holds media archives, recorded speech or health data, these marketplaces may be the better conversation. For most companies whose main asset is documented process, micro1, Mode and Mercor are the relevant programmes.
What can companies license?
The safest, most licensable material is documentation your firm wrote about how it works. Records about your clients, staff or suppliers need far more care. Start with the first list.
| Usually straightforward | Needs care (or stays out) |
|---|---|
| Standard operating procedures and process manuals | Customer and client records: CRM entries, account files, case files |
| Playbooks and runbooks: onboarding, escalation, sales and delivery methods | Email and chat (Slack, Teams), which mix process with personal data and confidential detail |
| Internal wikis and knowledge bases you wrote yourselves | Support tickets and workflow tickets that quote customers or name individuals |
| Project histories: plans, retrospectives, decision logs | Anything covered by an NDA or customer contract (MSAs often restrict reuse) |
| QA checklists, templates and review criteria | Third-party content in your knowledge base: licensed reports, vendor documentation, paid training material |
| Training material you created for your own staff | Employee data: HR files, performance reviews, recruitment records |
Why process documentation comes first
It is yours, so you have the right to license it. It usually holds little or no personal data, so the data-protection questions are smaller. And it is what the buyers say they want: a record of how the work gets done.
A practical way to scope it
- List the systems where your process knowledge lives (wiki, shared drives, ticketing, project tools) and how many years of history each holds.
- Mark each one “ours to license”, “contains client or personal data” or “covered by third-party terms”.
- Take the first group into the conversation with a programme. Raise the other two only after the legal checks in section 6.
How much could your data be worth to AI?
No one can tell you in advance. What we can show you is what the programmes publish, and what two programmes’ own calculators give for a typical firm.
What the programmes publish
| Programme | Published value | How the company is paid |
|---|---|---|
| micro1 | $100K–500K qualified partnership $500K–3M large-scale or multi-team $3M+ highly unique data | Data reviewed and approved within 30 days of transfer; paid within 30 days of invoice |
| Mode | $100K–$5M | Agreement sets the amount and schedule; no broker fee charged to the company; about 3 months from first conversation to payment |
| Mercor | A range per business type, from $10K–$300K to $100K–$1.5M | By wire on delivery of the anonymised dataset, typically within 2–4 weeks of signing; no fees or deductions |
| Protege | No range published | Revenue share under agreed partnership terms; percentage not published |
| Defined.ai | No range published | Licensing fees from marketplace sales; terms set per dataset and approved by the partner |
A worked example
For a firm with 40 full-time desk staff, with 8 years of records, here is what each calculator gives. They measure different things, so read them side by side, not as one range.
| Where the team is based | Mode’s previous public estimator | micro1’s calculator |
|---|---|---|
| United States | $190K–$390K | $270K–$550K |
| United Kingdom | $150K–$290K | $200K–$410K |
Mode: the formula Mode published in its estimator until October 2026, which it now keeps for earlier applications. It scales with desk staff, years of records and region. micro1: readings we took from micro1’s calculator (micro1.ai/data-payout) on 30 October 2026, at the nearest inputs we recorded (50 employees, 8 years).
These are indicative estimates, not offers. Each comes from one programme’s own calculator. Neither tells you what that programme, or any other, will offer your firm. Final offers depend on the records you can actually license, the programme’s review and the contract.
Try your own numbers with the calculator at trackimpact.io/lp/data-partnerships.
Does your firm qualify?
Size, location and language are the first filters. Where a programme does not publish a requirement, the table says so.
| Staff minimum | Regions | Language | Data types | |
|---|---|---|---|---|
| micro1 | 30+ full-time employees | US first, then other Western markets; demand strongest in US, UK, Canada | Mainly English | SOPs, knowledge bases, CRM, project histories, internal processes, code |
| Mode | Law firms 6+ Accounting, finance, medical 10+ Everyone else 30+ | Mainly US teams strongest; calculator also covers Canada, UK, Australia, EU | Not published | Email and calendar, documents, chat, call transcripts, CRM, billing |
| Mercor | Varies by business type, 5 to 50 staff | Varies; many “Anywhere”, some US only or US preferred | Not published | Workplace tools you connect (100+ supported) |
| Protege | Not published | Not published | Not published | Healthcare, video, audio and speech, spatial and physical, other domains; you must own or control the rights |
| Defined.ai | “No fixed minimum” | Not published | Not published | Audio, video, image, text, multimodal; in demand: cinematic video, anime, code, STEM |
Signs your firm is a strong fit
- An established company at or above the staff thresholds, with several years of operating history.
- Processes written down in one or more systems: a wiki, SOP library, ticketing tool or project tracker.
- Mainly English-language records.
- Clear ownership of the documentation, without customer contracts that forbid reuse.
Is it legal to license this data for AI training?
Often yes, for the right material and with the right checks. The main risks are your contracts, data-protection law, professional rules and weak de-identification. Work through this list with your counsel before sharing anything.
1. Rules specific to companies
- Law firms: legal privilege and professional conduct rules restrict client material heavily. Internal know-how and process are a very different question from client files.
- Accounting and tax firms: professional duties of client confidentiality apply to client files.
- Healthcare and anything near it: patient data is special-category data. Limit any conversation to non-clinical administration.
2. Customer contracts and NDAs
- Read your master services agreements, NDAs and data-processing agreements. Many limit how you may use information received from customers, including in anonymised form.
- Check supplier and partner agreements for the same restrictions.
3. Data-protection law
- UK GDPR and EU GDPR: personal data collected for one purpose (running an account, employing someone) generally can’t be reused for an incompatible one, such as AI training, without a lawful basis. This is the purpose-limitation principle. Properly anonymised data falls outside GDPR; pseudonymised data does not.
- US state privacy laws such as California’s CCPA bring their own notice, sale and sharing rules for personal information.
- Special-category data (health, for example) carries extra restrictions. Keep it out unless your counsel advises otherwise.
4. Third-party content
- Knowledge bases often hold material you licensed from someone else: research reports, vendor manuals, paid courses. You usually don’t have the right to license it on.
5. De-identification: questions to ask the programme
- Who de-identifies the data, using what method, and before or after it leaves your systems?
- Can you review a de-identified sample before anything is delivered?
- How do they handle names, client identifiers and free text that may reveal who someone is?
- What happens if personal data is found after delivery?
How does it work, and how long does it take?
Each programme publishes its own steps. Mode states that you don’t hand over passwords; ask the others how data leaves your systems.
| Programme | Published steps | Published timing |
|---|---|---|
| micro1 | Company profile and data manifest → review → call with a data partnership manager → contract → data transfer → approval → payment | Approval within 30 days of transfer; payment within 30 days of invoice |
| Mode | Estimate → non-binding letter of intent → final offer → your team signs in to the approved systems and Mode takes an agreed, filtered copy | About 3 months from first conversation to payment |
| Mercor | Scope → connect your tools → automated extraction and anonymisation → delivery → payout | Typically 2–4 weeks from signed agreement to payment |
| Protege | Discovery → revenue share agreement → Protege prepares the data (curation, quality control, masking) → data included in qualified deals | Says it often fulfils active opportunities “in under a month” |
| Defined.ai | Account creation → market research → data evaluation (quality, volume, metadata) → ethics and compliance check (licensing rights, legal, cybersecurity) | Not published |
What to prepare before the first call
- A short list of the systems that hold your process documentation, and the years of history in each.
- A view, even a rough one, of what is “ours to license” and what isn’t (section 3).
- A named internal owner, plus someone who can speak for legal or compliance.
- Your headcount, split between desk and non-desk staff, and where the team is based.
Questions to ask before signing
Programmes publish their ranges and steps, but much of what matters sits in the contract. Get these answered in writing.
- Exclusivity. Is the licence exclusive? Can you license the same material to anyone else?
- Onward use and resale. Can the programme resell or sublicense your data to other AI developers? To whom?
- Scope. Exactly which systems, records and date ranges are included, and what is excluded?
- De-identification and sample review. Who does it, how, and can you approve a sample before delivery?
- Retention and deletion. How long is the data kept, and can you require deletion at the end of the term?
- Liability and indemnity. Who is liable if personal or confidential data gets through, or a customer objects?
- Payment. The amount, the schedule, and any conditions or clawbacks, in writing. For revenue-share models: how use is measured and reported.
- Ownership. micro1 and Protege say companies keep ownership, and Mode licenses an agreed copy. Confirm the contract says so too.
A programme that answers these clearly is easier to work with. One that won’t answer them is telling you something too.
Frequently asked questions
Can I sell my company’s data to AI companies?
In most cases you would license an agreed copy of selected records rather than sell them outright. micro1 and Protege state that companies keep ownership. As of October 2026, micro1 and Mode run programmes for established companies, focused on internal process documentation.
How much is my company’s data worth to AI?
Programmes publish ranges: micro1 lists tiers from $100K–500K to $3M+, Mode lists $100K–$5M, and Mercor lists a range per business type from $10K to $1.5M. For a US company with 40 desk staff and 8 years of records, Mode’s previous public estimator gives an indicative $190K–$390K, and micro1’s own calculator gives $270K–$550K. Those are estimates, not offers.
What company data do AI companies buy?
Mainly records of how work gets done: SOPs, playbooks, wikis, project histories, QA checklists and templates. Customer records, email and chat can be in scope too, but need far more legal care. Marketplaces such as Protege and Defined.ai focus on health, media, audio, code and STEM data.
Is it legal to license company data for AI training under UK GDPR, GDPR or CCPA?
It can be. Material with no personal data, such as process documentation you own, raises the fewest issues. Personal data generally can’t be reused for AI training without a lawful basis, and customer contracts may forbid reuse. Get advice from your counsel; this guide is not legal advice.
Is the micro1 data partnership programme legitimate?
micro1 announced a $1bn, 12-month commitment to license company operational data on 9 October 2026, with capital from Citi and Hercules Capital, and publishes its tiers, eligibility and process. As with any counterparty, check the contract terms in section 8 before signing.
Is Mode’s data programme legitimate?
Mode (mode.inc) publishes its range, minimum sizes, process and how access works, and states that it charges companies no broker fee. Judge it, like any counterparty, on the written offer and contract.
micro1 or Mode: which is better for my company?
It depends on size, location and sector. micro1 looks for 30+ employees and is strongest in the US, UK and Canada. Mode looks for 30+ staff in most sectors, 10+ for accounting, finance and medical, and 6+ for law firms. Some companies qualify for both, and nothing on either programme’s public pages stops you applying to both.
How long does it take to get paid?
micro1 says it reviews and approves the data within 30 days of transfer and pays within 30 days of invoice. Mode describes about three months from first conversation to payment. Mercor says it typically pays within 2–4 weeks of the signed agreement.
Does TrackImpact see or handle my data?
No. We match companies to the programmes they qualify for. Programmes contract with you directly, and no data passes through TrackImpact. Data Partnerships is separate from the TrackImpact AI-adoption platform, and platform customer data is never part of it.
Kept up to date on our website
This guide is a snapshot dated 30 October 2026. The programmes change their terms often; these pages are updated when they do.
- Every programme, compared: micro1, Mode and Mercor side by side, updated as they change. trackimpact.io/data-licensing/programmes/
- Does my company qualify?: Minimums by sector and location, including the UK. trackimpact.io/data-licensing/guides/does-my-company-qualify/
- What is company data worth?: Published ranges, calculator readings and what moves the price. trackimpact.io/data-licensing/guides/what-is-company-data-worth/
- Questions to ask before you sign: A printable checklist. trackimpact.io/data-licensing/guides/questions-to-ask/
- Prepare a data inventory: The worksheet programmes ask for. trackimpact.io/data-licensing/guides/data-inventory/
- Is this offer legitimate?: Red flags, and three questions for any broker. trackimpact.io/data-licensing/guides/is-this-offer-legitimate/
- How we verify figures: Where every number comes from. trackimpact.io/data-licensing/methodology/
Check your fit in 60 seconds
There are now several programmes with different thresholds, regions and data types. Our fit check asks a few short questions about your firm and tells you which programmes you are likely to qualify for, with an indicative range and notes on your fit. We then send a short guide to scoping what to offer and the questions to ask before signing.
TrackImpact Data Partnerships fit checkFind out which programmes your firm is likely to qualify for and see an indicative range. About a minute.trackimpact.io/lp/data-partnerships →Not advice. This guide is general information. It isn’t legal, tax, financial, investment or other professional advice, and it isn’t a recommendation to license data or to join any programme. Figures are as each programme published them on the date shown; they can change, they aren’t offers, and nothing is guaranteed. Programmes decide who they accept, what they pay and on what terms. Check customer contracts, NDAs and data-protection obligations with your own advisers, and confirm everything with the programme, before sharing anything. To the extent the law allows, TrackImpact accepts no responsibility for decisions made on the basis of this guide, or for any programme’s terms, conduct or payments.
Disclosure. TrackImpact may receive a referral fee from programmes. TrackImpact is not affiliated with, employed by or acting for any programme named in this guide. Estimates are indicative and are not offers. This guide is not legal advice.
TrackImpact is a trading name of Flex Luthor Limited, registered in England and Wales (company number 05416368). Contact: contact@trackimpact.io
Sources, checked 30 October 2026
- micro1: $1bn commitment, Citi and Hercules Capital. micro1.ai/blog/micro1-commits-1b-to-enterprise-data-acquisition (9 October 2026)
- micro1: tiers, eligibility, process, payment timing, ownership. micro1.ai/data-partnerships
- micro1: payout calculator readings. micro1.ai/data-payout
- micro1: company referral reward and Defunct Data Partnership. micro1.ai/company-referral
- Mode: range, minimum sizes, process, access, ownership. data.mode.inc
- Mode: previous public estimator formula. data.mode.inc/pricing.js (retained by Mode for earlier applications)
- Mercor: process and payment. mercor.com/data
- Mercor: ranges by business type. data.mercor.com/refer
- Protege: funding. Business Wire release, January 2026; Lowenstein Sandler announcement of the $30M round led by a16z
- Protege: data-provider programmes, revenue share, process, ownership. withprotege.ai/data-provider/healthcare and /other-domains
- Defined.ai: data partner programme. defined.ai/partnerships and defined.ai/partnerships/ai-service-supplier
- Defined.ai: company background. defined.ai/llms.txt
Data licensing for accounting & tax firms
Yes, many established accounting and tax firms can, as of October 2026, as long as client files stay out. AI programmes are licensing records of how professional work gets done, and a firm’s own methodology, checklists, templates and workflow history are exactly that. Mode accepts accounting firms from 10 staff and publishes a range of $100K–$5M. micro1 looks for 30+ employees and publishes tiers from $100K–500K to $3M+. Client returns, workpapers and correspondence are a different matter: professional confidentiality, and in the US the tax-preparer disclosure rules, put them off-limits in practice.
| Usually straightforward | Needs care (or stays out) |
|---|---|
| Engagement and review checklists your firm wrote | Client files: returns, workpapers, ledgers, supporting documents |
| Close, reconciliation and review procedures | Client correspondence: email, portal messages, call notes |
| Tax-preparation workflows and templates (with no client data in them) | Anything under engagement letters or confidentiality terms |
| Internal methodology, technical notes and training | Tax return information (in the US, restricted by IRC section 7216) |
| Practice-management workflow history (job and task flows, stripped of client detail) | Third-party content: tax research subscriptions, CPD and paid training material |
| Quality-control and peer-review procedures | Staff data: HR files, appraisals, timesheets that identify people |
Worked example for a US firm with 12 desk staff and 10 years of records: Mode’s previous public estimator gives an indicative $170K–$340K (UK: $130K–$260K). micro1 asks for 30+ employees, so we don’t quote its calculator here. Estimates, not offers.
Can an accounting firm sell client data to AI companies?
Not realistically. Client files, returns and correspondence are covered by professional confidentiality and, in the US, by tax-preparer disclosure rules. What a firm can license is its own know-how: methodology, checklists, templates and workflow history with client detail removed.
Does a small accounting firm qualify?
Mode accepts accounting firms from 10 staff, lower than its usual 30. micro1 looks for 30+ employees, and Mercor lists accounting and tax firms from 5. Several years of documented process makes a firm a stronger fit.
How much could an accounting firm’s data be worth?
Programmes publish ranges from $100K up to $5M. For a US firm with 12 desk staff and 10 years of records Mode’s previous public estimator gives an indicative $170K–$340K (micro1 asks for 30+ employees). Those are estimates, not offers.
Does IRC section 7216 apply to licensing data for AI?
If the data includes tax return information, section 7216’s restrictions on use and disclosure by return preparers apply. The safe course is to keep return information out of anything you license and to take advice before including anything derived from it.
Data licensing for marketing agencies
Yes, many established agencies can, as of October 2026. AI programmes are licensing records of how work gets done, and an agency’s own playbooks, briefing templates, QA and approval workflows and project histories fit that well. Mode looks for 30+ staff (mainly US teams) and publishes a range of $100K–$5M; micro1 looks for 30+ employees and publishes tiers from $100K–500K to $3M+. The line to watch is client material: briefs, creative work and campaign data are usually covered by client contracts, and creative work is often assigned to the client.
| Usually straightforward | Needs care (or stays out) |
|---|---|
| Agency playbooks and delivery methods | Client briefs, strategies and creative work (often client-owned under the MSA) |
| Briefing and proposal templates (without client detail) | Campaign and ad-account data belonging to clients |
| QA, approval and sign-off workflows | Client correspondence: email, Slack Connect, feedback rounds |
| Project histories in your project tool, stripped of client detail | Anything under client NDAs or confidentiality clauses |
| Onboarding and training material you created | Licensed assets: stock imagery, fonts, music, paid research |
| Internal process docs: resourcing, estimating, retrospectives | Staff and freelancer data |
Worked example for a US firm with 25 desk staff and 8 years of records: Mode’s previous public estimator gives an indicative $170K–$350K (UK: $130K–$260K). micro1 asks for 30+ employees, so we don’t quote its calculator here. Estimates, not offers.
Can a marketing agency license its data to AI?
Often, for its own process material: playbooks, templates, QA workflows and project histories with client detail removed. Client briefs, creative work and campaign data are usually covered by client contracts and are not the agency’s to license.
Who owns agency creative work for licensing purposes?
It depends on the contract. Many agency MSAs assign IP in deliverables to the client. Check each MSA before treating any client work as licensable.
How much could an agency’s data be worth?
Programmes publish ranges from $100K up to $5M. For a US agency with 25 desk staff and 8 years of records Mode’s previous public estimator gives an indicative $170K–$350K (micro1 asks for 30+ employees). Those are estimates, not offers.
Does an agency need client permission?
For anything containing client material, usually yes. For your own process documentation with client detail removed, usually not, but check your contracts with counsel first.
Data licensing for recruiting & staffing firms
Yes, many established recruiting and staffing firms can, as of October 2026, if candidate data stays out. AI programmes are licensing records of how work gets done, and a firm’s screening playbooks, interview guides, placement workflows and compliance procedures fit that well. Mode looks for 30+ staff (mainly US teams) and publishes a range of $100K–$5M; micro1 looks for 30+ employees and publishes tiers from $100K–500K to $3M+. Candidate CVs, notes and assessments are personal data collected for recruitment, and reusing them for AI training is a different purpose.
| Usually straightforward | Needs care (or stays out) |
|---|---|
| Screening and qualification playbooks | Candidate CVs, profiles and notes (personal data) |
| Interview guides and scorecards your firm wrote (blank) | Assessment results and references |
| Placement workflows and pipeline stage definitions | Diversity, health or right-to-work data (special care or special category) |
| Compliance and onboarding procedures | Client job specs and fee terms (client-confidential) |
| Training material for recruiters | Email and messaging history with candidates and clients |
| Process history from your ATS or CRM, stripped of personal data | Third-party data: job-board and data-provider content under licence |
Worked example for a US firm with 30 desk staff and 10 non-desk staff and 8 years of records: Mode’s previous public estimator gives an indicative $180K–$360K (UK: $140K–$270K), and micro1’s own calculator gives $270K–$550K (UK: $200K–$410K). Estimates, not offers.
Can a recruitment agency sell candidate data to AI companies?
Not realistically. Candidate data was collected for recruitment, and reusing it for AI training is a different purpose under UK and EU GDPR; some of it is special-category data. What a firm can license is its own process: playbooks, guides and workflows with personal data removed.
What can a staffing firm license?
Screening and qualification playbooks, blank interview guides and scorecards, placement workflows, compliance procedures, training material, and process history stripped of personal data.
How much could a recruiting firm’s data be worth?
Programmes publish ranges from $100K up to $5M. For a US firm with 30 desk staff, 10 non-desk staff and 8 years of records Mode’s previous public estimator gives an indicative $180K–$360K and micro1’s calculator gives $270K–$550K. Those are estimates, not offers.
Do non-desk staff count?
Mode’s previous public estimator counts non-desk staff at a much lower weight than desk staff. micro1’s calculator asks for total employees.
Data licensing for consulting firms
Yes, many established consulting firms can, as of October 2026. AI programmes are licensing records of how professional work gets done, and a firm’s own methodologies, frameworks, templates and engagement playbooks are exactly that. Mode looks for 30+ staff (mainly US teams) and publishes a range of $100K–$5M; micro1 looks for 30+ employees and publishes tiers from $100K–500K to $3M+. Client deliverables and client data are usually covered by engagement terms, and deliverables are often client-owned.
| Usually straightforward | Needs care (or stays out) |
|---|---|
| Methodologies and frameworks your firm developed | Client deliverables (often client-owned under engagement terms) |
| Engagement playbooks and delivery methods | Client data used or produced during engagements |
| Templates: proposals, workplans, analyses (without client detail) | Proposals and case material that name clients |
| Internal knowledge base and training material | Anything under client NDAs or confidentiality clauses |
| Quality-review and sign-off procedures | Third-party research: analyst reports and data subscriptions |
| Project histories stripped of client detail | Staff data: appraisals, utilisation records that identify people |
Worked example for a US firm with 30 desk staff and 8 years of records: Mode’s previous public estimator gives an indicative $180K–$360K (UK: $140K–$270K), and micro1’s own calculator gives $230K–$450K (UK: $170K–$330K). Estimates, not offers.
Can a consulting firm license its methodology to AI?
Often, yes. A firm’s own frameworks, playbooks and templates are typically its property. Client deliverables and client data are usually covered by engagement terms and are not the firm’s to license.
Who owns consulting deliverables?
It depends on the engagement terms. Many make deliverables client property. Check your standard terms and any client-specific contracts first.
How much could a consulting firm’s data be worth?
Programmes publish ranges from $100K up to $5M. For a US firm with 30 desk staff and 8 years of records Mode’s previous public estimator gives an indicative $180K–$360K and micro1’s calculator gives $230K–$450K. Those are estimates, not offers.
Can boutique firms qualify?
Mode and micro1 both look for 30+ staff. Mercor lists consulting and advisory firms from 10 employees, so a boutique below 30 has at least one published option.
Data licensing for software & it firms
Yes, many established software and IT firms can, as of October 2026. AI programmes are licensing records of how work gets done, and runbooks, incident reviews, internal engineering docs and ticket workflows fit that well. Mode looks for 30+ staff (mainly US teams) and publishes a range of $100K–$5M; micro1 looks for 30+ employees and publishes tiers from $100K–500K to $3M+. Watch three things: customer data in tickets and logs, code your clients own, and secrets left in docs.
| Usually straightforward | Needs care (or stays out) |
|---|---|
| Runbooks and operational procedures | Customer tickets and logs that contain customer or user data |
| Incident reviews and postmortems (de-identified) | Client-owned code (common in IT services and dev shops) |
| Internal engineering docs, architecture notes, standards | Credentials, keys and secrets left in docs or tickets |
| Ticket workflows and support processes, stripped of customer detail | Anything under customer contracts, DPAs or NDAs |
| Onboarding and training material | Third-party code and docs under licences that restrict reuse |
| Your own code, as a separate decision (it is core IP) | Security reports and vulnerability details |
Worked example for a US firm with 40 desk staff and 6 years of records: Mode’s previous public estimator gives an indicative $180K–$360K (UK: $140K–$270K), and micro1’s own calculator gives $240K–$460K (UK: $180K–$350K). Estimates, not offers.
Can a software company license its internal docs to AI?
Often, yes: runbooks, de-identified postmortems, engineering docs and ticket workflows with customer data removed. Customer data you process under a DPA is not yours to license.
Should we license our source code?
That is a separate decision from licensing process documentation, because code is usually core IP. Client-owned code and open-source licence obligations also need checking. Take advice before including any code.
How much could a software firm’s data be worth?
Programmes publish ranges from $100K up to $5M. For a US firm with 40 desk staff and 6 years of records Mode’s previous public estimator gives an indicative $180K–$360K and micro1’s calculator gives $240K–$460K. Those are estimates, not offers.
What should we scan for before sharing?
Customer and user personal data, credentials and API keys, internal hostnames and infrastructure detail, and security or vulnerability information.
Data licensing for real estate & property firms
Yes, many established real estate and property firms can, as of October 2026. AI programmes are licensing records of how work gets done, and transaction checklists, property-management procedures, maintenance and leasing workflows fit that well. Mode looks for 30+ staff (mainly US teams) and publishes a range of $100K–$5M; micro1 looks for 30+ employees and publishes tiers from $100K–500K to $3M+. Client, tenant and buyer data stays out, and listing data from an MLS is licensed to you, not yours to license on.
| Usually straightforward | Needs care (or stays out) |
|---|---|
| Transaction checklists and closing procedures | Client, buyer and tenant personal data |
| Property-management procedures | Financial information: mortgage, income and payment details |
| Maintenance and work-order workflows, stripped of tenant detail | Contracts, leases and offers naming parties |
| Leasing and onboarding processes | MLS listing data (licensed to you under MLS rules) |
| Compliance procedures your firm wrote | Email and messaging history with clients and tenants |
| Training material for agents and managers | Third-party data: valuation tools and data subscriptions |
Worked example for a US firm with 20 desk staff and 5 non-desk staff and 10 years of records: Mode’s previous public estimator gives an indicative $180K–$360K (UK: $140K–$270K). micro1 asks for 30+ employees, so we don’t quote its calculator here. Estimates, not offers.
Can a real estate brokerage license its data to AI?
Often, for its own process material: transaction checklists, procedures and workflows with client detail removed. Client data and MLS listing data are not the brokerage’s to license.
Can we include MLS listing data?
No. MLS listing data is licensed to members under MLS rules that typically restrict redistribution.
How much could a property firm’s data be worth?
Programmes publish ranges from $100K up to $5M. For a US firm with 20 desk staff, 5 non-desk staff and 10 years of records Mode’s previous public estimator gives an indicative $180K–$360K (micro1 asks for 30+ employees). Those are estimates, not offers.
Do property-management firms qualify?
If they meet the staff thresholds: 30+ for Mode and micro1. Several years of documented procedures make a firm a stronger fit.
