Sell Agency DataGet a quote

For agency founders and partners

Sell Agency Data

License your briefs, revision rounds and timesheets to AI training-data buyers, with the client's identity taken out first. The finished work usually stays the client's. The record of how you made it is often yours.

Get a quoteWhat you own vs the client

One job record, from your project tool to a paid license

1 Raw job record

Job 0412Spring launch page

Client
Harbor & Pine Outdoor
Account lead
Priya Shah
Brief v1
Lead the hero on price
Brief v3
Lead on durability, cut the promo
  • R1Layout sent back
  • R2Copy changes
  • R3Approved
Design 31.5hDev 18hCopyPM

Logged against the job69.75 h

2 Client removed

Job 0412Launch page

Client
removed
Account lead
role_am_02
Brief v1
Lead the hero on price
Brief v3
Lead on durability, cut the promo
  • R1Layout sent back
  • R2Copy changes
  • R3Approved
Design 31.5hDev 18hCopyPM

Names and brand gone69.75 h kept

3 You choose the buyer

  • micro1
  • Scale AI
  • Nyne
  • Labelbox Prism

4 Paid under a license

30 days

micro1 pays the whole amount within a month of accepting the data. You license; you do not sell title.

Invented example. The client, the job and the people in step 1 are made up.

The records that sell are the ones that show the work

Labs training AI agents to run projects need the messy middle: the brief that changed three times, the feedback that sent a layout back, the hours logged against each task. None of the buyers asks for a portfolio.

Revision rounds

Feedback, the response and the next version, in order. Agencies hold this in quantity and almost nobody else does, which is why it carries the most interest.

  • Feedback
  • Response
  • Next version
  • Sign-off

Briefs and change requests

The starting intent and every dated amendment. Remove the client's name, product and anything a reader could trace back.

Timesheets

Hours by task, role and phase over years. Structured, usually the agency's own record, with staff names taken out.

QA and sign-off

Checklists, bug tickets and approval steps: how a team decides something is done.

Process documents

Onboarding guides, SOPs and estimating templates. Plainly yours, and on micro1's published list.

Recruiting agencies: the people stay out

Resumes, interview notes and salary expectations are personal information about candidates who never agreed to AI training. Role specs without the client, stage timings, outreach templates and screening rubrics may go, depending on your candidate privacy notice and the states you recruit in.

What a recruiting agency can license

Ownership

The client owns the deliverable. You often own the trail.

Start with the MSA, not the archive. Once the invoice clears, the IP assignment clause in most agency contracts hands the finished work to whoever paid for it. Timesheets, internal QA and your own process documents are often the agency's. Some contracts define confidential information broadly enough to catch the brief itself.

The clause-by-clause checklist

Usually the client's

  • Final deliverables and source files
  • Brand assets, data and research
  • Anything the contract calls confidential
  • Their customers' personal data

Often yours, if the contract agrees

  • Timesheets and resourcing records
  • Process docs, templates and SOPs
  • Revision logs and QA notes, client removed
  • Briefs, once nothing identifies the client

Size

Most buyers start counting at 20 staff

Nyne asks for the least: twenty W-2 employees on a US payroll and three years of trading. micro1 asks for 30 full-time staff and English-language records. Scale AI prefers 40 but judges each case. Labelbox Prism prints no size rule at all.

See which ranges your headcount fits

Printed staff floors

  • 20NyneUS payroll, three years trading
  • 30micro130+ full-time staff
  • 40Scale AI40+ preferred, case by case
  • AnyLabelbox PrismNo general rule printed

Buyers

Four buyers print their terms, and three name agency-style records

All four sell training data on to AI labs. micro1 names process documentation and SOPs, Scale AI names Jira and Confluence work, Nyne names Asana and Jira projects. Mercor buys company data too, but we print nothing about its terms.

Payment timing, exclusivity and referral terms · Terms changelog

  • micro1$100K to $3M+SOPs, planning and operations records, knowledge bases
  • Scale AI$10K to $1M+Jira and Confluence records, files with metadata, email, calendars
  • Nyne$100K+Asana and Jira project records, CRM, invoices and estimates
  • Labelbox Prism~$50K to $1M+Nothing agency-specific; cross-system interaction records

Our interest, declared: micro1 and Scale AI pay introducers, so a signed deal with either after our introduction can earn us a fee. No fee changes what we print here. Scale AI calls its range illustrative. Independent site, not affiliated with any company named.

Clients

Tell clients before they hear it elsewhere

If anything in the set came from their project, even with their name removed, the safer course is to say so. Read the confidentiality clause first: if it covers "all information relating to the project", even a de-identified brief may need consent.

How to write the note to clients

The question a client asks is rarely "was it legal?" It is "why did we hear about this from someone else?"

  • What wentTimesheets, revision logs
  • What did notDeliverables, their data
  • HowThe de-identification used

Published ranges sit at the top of the market

No agency's fee is public. These are the ranges buyers advertise, and they are marketing, not offers. Until a buyer has seen a sample, a mid-sized agency's archive usually sits well below the top of any of them.

  • $100K to $3M+micro1, three tiers per company
  • $100K+Nyne, described as common
  • ~$50K to $1M+Labelbox Prism, fee plus revenue share
  • $10K to $1M+Scale AI, illustrative only

Reported, not published. Google bid $10 million for Spirit Airlines' de-identified business records, with a court still to approve it (Reuters). Labeling startups were said to be paying HVAC contractors around $150,000 each (The Information). A founder closing a crypto startup got about $25,000 for its code (MIT Technology Review).

Each figure is that company's own advertised range, and none of them is an offer. Try the worth estimate

Where these figures come from

From first call to a paid license, in five steps

Each buyer runs these in roughly this order. Nothing moves until you have signed.

  1. 1

    Eligibility

    The buyer checks size, years, country and record types against its published rules.

  2. 2

    Scope and samples

    You agree which systems and date ranges are in. micro1 lets companies look over representative samples drawn from their records.

  3. 3

    De-identification

    Scale AI strips identities on your own systems before any export and asks you to sign off each package. Elsewhere the location goes unstated.

  4. 4

    The license

    Scope, rights, exclusivity and acceptance criteria. Labelbox Prism licenses are exclusive by default. Your own lawyer reads it.

  5. 5

    Acceptance and payment

    micro1 pays the full sum within 30 days of accepting the data. Labelbox says Prism deals can go from first call to payment in under three weeks.

How we make the introduction

What agency founders ask before they license

Can a marketing agency sell its data to AI companies?

It can, provided every record in the deal is its own or covered by a right to license it. For most agencies that means process records (briefs stripped of client identity, revision histories, timesheets, QA checklists) rather than the finished work, which the client usually owns.

Can we use client work to train AI or license it?

Usually not without the client agreeing. Most MSAs assign the work product to the client once the invoice is paid, so the campaign, the site build and the source files are theirs to license, not yours. Some confidentiality clauses reach the brief itself. See client-owned vs agency-owned.

Do we hand over client deliverables?

Usually not. Most MSAs assign deliverables to the client on payment. What buyers want from an agency is how work got made, which is often the agency's own record.

Will the buyer see client names?

They should not. All four promise de-identification, and Scale AI does it on your side of the fence. Even so, take out client names and personal details yourself before a sample goes anywhere; a buyer's clean-up is a backstop.

How big does the agency need to be?

Nyne sets the lowest printed bar: a US firm, three years in business, with twenty or more W-2 employees on the payroll full time. It rises to 30 full-time staff at micro1 and a preferred 40 at Scale AI. Labelbox Prism publishes no size rule.

What does an agency archive sell for?

Nobody publishes agency deals. Buyers publish ranges (from around $50K at Labelbox Prism to $3M+ at the top of micro1's tiers), and reported deals are often far smaller. The worth estimate shows which published ranges your agency's size fits.

Is it legal for a recruiting agency to license candidate records?

Candidate personal data is the hard part: resumes, contact details and interview notes are personal information about people who agreed to be considered for a job, not to train a model. The process with the people taken out can sometimes go. See recruiting agency data for what can and cannot.

Get a quote

Describe the agency and where its history is kept, in a few lines. Your answers get compared with what each buyer says it requires, and the email back names any that match. No buyer hears of you until you agree.

Keep records, files, samples and client names out of this form. A description of what the agency holds is plenty.

Step 1 of 3

Where is the company based?

Use the main office. Several buyers limit where they buy from, so location matters.

Get a quote