The short answer: Most approved deals for established companies land between about $100,000 and $2 million, according to ranges buyers publish, with many around $200,000 to $300,000. Smaller companies start nearer $20,000, and closed startups have sold their archives for $10,000 to $100,000. Your price depends on how rare the records are, how many continuous years you have, whether outcomes are recorded, how well records link across systems and how clean your rights are.
What buyers have actually paid
Most company data deals are private, but enough has been reported to sketch the market. In August 2026, Google won a bankruptcy auction for Spirit Airlines’ internal data with a $10 million bid. The package covered about 100 million emails, 500 million Microsoft Teams messages, operations records and code, with personal data and passenger records excluded. Court approval was still pending in October 2026, with a sale hearing set for November 6.
Forbes reported in April 2026 that one wind-down firm had arranged about 100 sales of closed startups’ Slack, email and Jira archives in a year, paying $10,000 to $100,000 per company.
For operating companies, buyers and brokers publish ranges rather than averages. Most approved deals land between $100,000 and $2 million, and buyers who mention a typical deal put it around $200,000 to $300,000. One AI data company has published program tiers of $100,000 to $1 million or more for companies with 30 or more staff.
Publisher deals sit far above that. Reddit reportedly earns about $60 million a year from Google, and News Corp’s agreement with OpenAI was reported at more than $250 million over five years. Those are platforms with huge, clean archives, so treat them as the top of the market, not a benchmark for one company. Company names here only report public deals and don’t imply any partnership or current buyer interest.
The six things that set your price
- Rarity. Could the buyer get something similar elsewhere? Specialist work in a narrow field beats a larger generic archive.
- Continuous history. Years on the same systems. A migration that dropped comments or links resets the clock.
- Outcomes. Resolved, won, lost, approved, rejected, paid. A record of what happened next turns activity into a lesson a model can learn from.
- Linked records. A ticket tied to its chat thread, the code change and the CRM note is worth more than the three apart.
- Volume, as a threshold. Enough data gets a buyer to look. Past that point, more of the same adds less than you’d expect.
- Clean rights. Data you can show you own and may license, with personal information removable, closes faster and holds its price.
What different kinds of data tend to earn
- Support tickets and chats with resolutions are among the most requested, because they show how a task gets finished.
- CRM and sales history with won and lost reasons is valuable, especially when it links to emails and calls.
- Code with its review history, meaning pull requests, comments and linked tickets, is in strong demand for coding agents.
- Expert documents such as inspection reports, root-cause analyses and underwriting notes are scarce and price high per document.
- Email and team chat alone are worth less than owners expect, and much more when they link to the records they discuss.
- Finance and operations records are useful when they show decisions and results. They need the most care around personal and client data.
- Web analytics, marketing lists and contact lists are worth little to these buyers, or can’t be licensed at all.
How the money is paid
Most deals today are a one-time license fee for a copy of historical records, often paid in parts: some on signature and the rest on delivery and acceptance. Installments on one archive are still a one-time deal, not recurring revenue.
Operating companies can sometimes add refresh payments for new records delivered each year. Some buyers offer part of the price as credits, compute or access to a model. Those can be worth a lot, but compare them with cash carefully.
Exclusivity is the single biggest lever. A buyer who wants your data to itself for around 24 months will usually pay more for it. Our guide to exclusive and non-exclusive licenses explains the trade-off.
How to estimate your own range
- Count the people who work at a computer every day, now or at your peak.
- Note how many continuous years each main system holds, and when you last switched tools.
- List which systems record outcomes, not just activity.
- Separate records you own from records you hold for clients.
- Put those numbers into our estimator, then ask for a free valuation for a real range.
Be careful with any number you’re quoted
Nobody can give a firm price before a buyer has seen a description and, usually, a small sample you’ve approved. Offers for the same data can differ several times over between buyers, so one offer is not a market price.
Be wary of anyone who asks for a fee, or for raw data, before a signed agreement.
Common questions
Is there a price per gigabyte or per record?
Not for most company data. Volume decides whether a buyer looks, but price follows rarity, history, outcomes, linked records and rights. Some buyers quote per document for expert material, and those quotes vary widely.
Can a small company earn anything?
Sometimes. Smaller companies with years of specialist records can start from around $20,000, and closed startups have sold archives for $10,000 to $100,000. Below about 25 people at desks, fewer buyers are interested.
Does a bigger company always earn more?
Not always. A 40-person firm with ten years of one rare workflow can out-earn a 400-person company with three years of generic email.
Can we license the same data again?
Often, either after an exclusive term ends or under non-exclusive terms. Plan renewals and refreshes when you sign the first license.
Sources
- Bloomberg Law, “Google Aims to Boost AI With Purchase of Spirit Airlines Data”, Aug 2026
- TIME, “Why Google Bid $10 Million for a Failed Airline’s Data”, Aug 25, 2026
- Forbes, “AI’s New Training Data: Your Old Work Slacks And Emails”, Apr 16, 2026
- Reuters, reporting on Reddit’s content licensing deal with Google, Feb 2024
- Wall Street Journal, reporting on News Corp’s agreement with OpenAI, May 2024
Want to know where your company stands? The free valuation takes about ten minutes and needs no files.
Talkbase is not a law firm. This guide summarizes public information so you can frame questions for your counsel. It is not legal, tax or financial advice.