Data licensing guide
Data licensing guide

What is my company's data worth to AI companies?

By Ade Molajo · Updated · How we check figures

Quick answer

The programmes that buy company records publish ranges from $20K to $5M. micro1 publishes tiers from $100K–500K to $2M+, Mode $100K–$5M, and Mercor a range per business type, from $20K–$250K to $100K–$1.5M. Where you land depends on years of history, how complete and linked the records are, how many systems are in scope, and whether you can show the right to license them. Nobody can tell you your figure until a programme reviews you.

$20K–$5M
published range across the three programmes
$203K–$377K
micro1's calculator for a 30-person US firm with 5 years of records
5
things that move the price
0
programmes that guarantee a figure

01What the programmes publish

Three programmes currently publish what they pay established companies for operating records. These are their own figures, checked on their own sites.

Published company ranges on one $0–$5M scale: micro1 $100K to $2M and up, Mode $100K to $5M, Mercor $20K to $1.5M depending on business type.$0$1M$2M$3M$4M$5Mmicro1$100K–500K · $500K–2M · $2M+micro1.ai/data-partnerships · 30 Oct 2026Mode$100K–$5M per companydata.mode.inc · 30 Oct 2026Mercor$20K–$1.5M, by business typedata.mercor.com/refer · 30 Oct 2026
Each programme’s own published range, on one scale. Ranges, not offers.
ProgrammePublishedWhat decides where you land
micro1Tiers: $100K–500K, $500K–2M, $2M+Dataset size, workflow complexity, quality, domain expertise, uniqueness
Mode$100K–$5M per companySystems and dates in scope, years of history, volume and quality, ownership and access terms
MercorA range per business type, from $20K–$250K (call centres) to $100K–$1.5M (manufacturers)The size and completeness of your records

These are ranges, not offers. Every programme prices each company individually after reviewing its records.

02What micro1's own calculator says

micro1 has a public calculator. We took readings for a US company with five years of records at nine sizes, so you can see how its estimates move with headcount.

micro1’s own calculator, US company with 5 years of records, by headcount: estimates rise from about $203K–$377K at 30 employees to about $1.96M–$4.54M at 1,000. Readings taken 30 October 2026; indicative only.$0$500K$1M$1.5M$2M$2.5M$3M$3.5M$4M$4.5M$5M30 employees$203K–$376.8K50 employees$239.2K–$462.6K75 employees$284.4K–$570K100 employees$329.6K–$677.4K150 employees$420.1K–$892.1K200 employees$510.5K–$1.107M300 employees$691.4K–$1.536M500 employees$1.053M–$2.395M1,000 employees$1.958M–$4.542M
Readings from micro1’s calculator (micro1.ai/data-payout), US, 5 years of records, 30 October 2026. Indicative, not offers.

The estimate roughly scales with headcount, and micro1's calculator puts larger firms above its "$2M+" top tier. Its estimates for Canada and Europe come out lower, and for other markets lower again. micro1 says the calculator is indicative, and so do we. Our fit check shows micro1's reading and Mode's estimator side by side.

03Mercor's ranges: the most useful benchmark

Mercor is the only programme that publishes a range for each kind of business, and says what each range depends on. Even if you apply elsewhere, it's the closest thing to a price list.

Business typeMercor's published rangeWhat the figure depends on
Law firm$50K–$1.25MMatter size and completeness
Accounting and tax firm$25K–$865KClient count and filing seasons
Consulting and advisory firm$150K–$300KVolume and depth of engagements
Construction company$115K–$1MProject count and imagery
Manufacturing and industrial$100K–$1.5MRecord types and history
Software company$50K–$500KMore for large codebases or exclusive terms

Mercor's listings, checked 30 October 2026. Ranges, not offers. See all 24 business types.

04Five things that move the price

Five things that move what a company is offered: years of history, completeness of each matter or project, records that link together, how many systems are in scope, and proof of the right to license.01YearsLongerhistory02CompleteWholematters,projects03LinkedrecordsTicket tocode to doc04SystemsMore in scope05Right tolicenseProof you can
From what the programmes say drives price, including Mercor’s per-type listings. None publishes a formula.
  1. Years of history. A longer record of the same work is worth more. Mercor's listings ask for five or more years of closed matters, several filing seasons, or three or more years of commits.
  2. Completeness. Whole matters, whole projects and whole engagements, from start to sign-off, beat fragments.
  3. Records that link together. A ticket linked to the code change and the document it produced shows the full chain of work.
  4. How many systems are in scope. Email, documents, tickets and CRM together tell more than any one alone.
  5. Proof of the right to license. Records you clearly own, or that clients have agreed to, are worth more than ones with open questions.

05What buyers won't pay for

Mercor's listings spell out what isn't a fit. Across them:

  • Templates or marketing material on their own
  • Active matters or live projects
  • Logs without the case or context around them
  • Open-source copies, forks or AI-written code
  • Anything client contracts restrict

06How to raise the value

  • Lead with complete, closed work: finished matters, completed projects, shipped releases.
  • List every system and how many years each covers. Programmes price what they can see.
  • Check client contracts early, so you know what's in scope before you're asked.
  • Ask whether exclusivity changes the price. Mercor's software listing says exclusive terms can lift it.

Calculators are indicative. The real number is the programme's offer, after it reviews your records.

07The public deals, and why they're a different market

You'll see big numbers quoted. They're real, but they're a different kind of deal:

DealReported valueWhat was licensed
Reddit and Google (2024)About $60M a year, reported by ReutersPublic forum posts at internet scale
Reddit's data licensing contracts$203M in total contract value, in its IPO filingThe same, across buyers
News Corp and OpenAI (2024)Over $250M across five years, reported by The Wall Street JournalNews archives and content
Shutterstock (2023)About $104M of AI licensing revenue, reportedStock images and video
Spirit Airlines (2026)Google won the auction with a $10M bid; court approval was pending at the time of writingA bankrupt airline's operating data

The first four are public or published content, licensed by platforms with millions of contributors. They show that AI companies pay serious money for data, not what a 50-person firm's records are worth. Spirit is the closest to a company's own operating records, and it drew objections over the employee data in it. See is it legal?

08How to find out your figure

  1. Check you qualify: size, history and location by programme.
  2. List your systems and how many years each covers.
  3. Apply to more than one programme and compare the offers. Many companies can apply to all three.

Our free fit check does steps 1 and 2 in two minutes. The programme decides.

09Questions people ask

How much is my company data worth?

The programmes publish ranges, not prices: $20K to $5M across the three we track. Your figure depends on the programme's review of your records.

Is there a calculator?

micro1 and Mode both have one on their sites. Both say the result is indicative. Our fit check shows both side by side.

Why are the ranges so wide?

Because the records vary so much. Ten years of structured tickets from a specialist firm is a different asset from two years of loose files.

What did Reddit or News Corp get for their data?

Those are reported deals for public content at internet scale, not company operating records. They show the market is real, not what a 50-person firm will be paid.

Sources

Check which programmes you’re likely to matchA few short questions, about a minute. Free, no sign-up. The programme decides who qualifies.trackimpact.io/data →