Which buyers of company data publish the highest payout ranges?

Every published dollar range, highest top figure first

This table lists every buyer in the index that publishes a whole-company dollar figure, ordered by the top of its range. It comes straight from the index, so it changes when a row does. The order is the companies’ own claims, not our view of who pays most: a higher ceiling says nothing about what your company would be offered. Each name links to the full record and its sources. For how these ranges compare with written offers and reported closed deals, read what company data is actually worth.

Not in the table: buyers whose published figure is not a whole-company dollar range. pre.dev prices per million tokens, Troveo says typical full-company deals start at six figures, and Nyne says partnerships are commonly worth $100K or more. The methodology explains which figures count.

How to read a published range

Five kinds of figure sit behind the word “range”, and they are not the same:

• A floor and a ceiling. Handshake AI publishes $100K to $4M+ and Mode $100K to $5M, with Mode marking its calculator as indicative only.
• Tiers. micro1 publishes $100K to 500K, $500K to 2M and $2M+, and elsewhere says payments typically range from $150K to over $2M.
• A ceiling only. Turing publishes a single payout of up to $1M, Ooak Data up to $1M with closed deals from $20K, and SourceX up to $4M, which it calls an upper bound for the largest enterprise datasets. 3 buyers publish a ceiling with no floor.
• Estimator output. Telegraph Lab shows $385K to $895K for its default inputs, and Actual Data $285,896 to $604,463. A default estimator output is not a range the company commits to, so it only counts when the company also publishes a range.
• Figures cited from elsewhere. BuyMyDeadStartup cites $10,000 to $100,000 from press reports rather than its own deals, and Corpus Bridge cites a Forbes figure of about $10,000 to several hundred thousand dollars for ordinary companies.

Two companies, one set of deal figures

Telegraph Lab and Regulr show the same three deals: $650K for a freight broker, $450K for an engineering firm and $350K for an accounting firm. Telegraph Lab lists them as deals it paid, while Regulr says its examples have not been independently verified. Count them once.

The pattern repeats elsewhere. Grepped and Smelted publish the same $20K to $5M headline, and Actual Data’s requirements block matches micro1’s partnership posting word for word. When two sites publish the same figure, you have one data point.

Where the median sits

Across the 25 buyers that publish a whole-company dollar range, the median published range is $100K to $2M, and 21 of them publish a ceiling of $1M or more. The ceilings cluster at round numbers, which says more about how landing pages are written than about any deal. The floors say more: most sit at $100K, the lowest published floor is $10K, and Ooak Data is the only buyer whose floor is described as its smallest closed deal.

A broker’s range is its buyer’s range. Korra publishes $200K to $2M typical and says its fee is “agreed in writing before you sign”, without saying who pays it. Which way the fee runs matters as much as the range, and the payout model guide sorts the buyers by it.

Buyers that publish no range

22 of the 51 buyers publish no range. Among the established buyers, Appen, Mercor, Centific, Halluminate and Protege publish none. Mercor says compensation is based on data volume, the tools you connect and the depth of your data. Centific gives a written estimate after a first review. Appen agrees terms in writing before extraction begins. No range is not a lower offer. It is a company that has not said.

What moves a real figure, in the buyers’ own words

• Volume, tools and depth. Mercor names all three. Handshake AI’s figure “depends on the data”, and Scale AI’s illustrative value scales “with cadence”, meaning how often you deliver.
• Years of history. Handshake AI asks for 3+ years in operation, Nyne for 3+ years of operational workflows, and Mode says companies with several years of records they own are the strongest fit.
• Recurring deliveries. Centific says most partners move to a quarterly refresh, where a one-time sale becomes a recurring line, and Scale designs partnerships for recurring revenue.
• Exclusivity. SourceX licenses exclusively for AI training. Labelbox Prism structures licenses as exclusive by default. pre.dev is non-exclusive by default, so a repo can pay more than once, and SuperAnnotate pays on delivery and on each relicense under a time-limited license.
• A review before a number. Ooak Data: “The estimator gives a range; the audit gives a figure.” Longitude Labs quotes the whole archive after reviewing two or three samples.

Next step

The value estimator applies these published ranges to your company’s size and years, and the size guide shows which buyers’ published minimums you meet. For a closer estimate with the buyers that fit and the terms to watch, use the free value check. If you already have a figure in writing, the offer checklist covers the clauses that decide what the figure is worth.