How do data licensing offers work?

Most data licensing offers come in two stages: an indicative range from a description of your company, then a firm price once a buyer has inspected your records and the terms are written down. The first number shows interest; the second says what you would be paid, for which records, and when.

What is the difference between an indicative offer and a firm offer?

An indicative offer is a range worked out from what you have told the buyer: headcount, years in business, the systems you use. Avelence says its result is “not annual revenue, a single-license quote, a minimum payment, or a buyer offer”, and Nyne labels its figure “indicative and not an offer”. At a default of 100 employees and 10 years, Handshake AI’s and Nyne’s estimators both showed $383,083 to $896,969 and the calculator of Telegraph Lab (affiliated with this site) showed $385K to $895K, as of October 2026 (Handshake AI, Telegraph Lab). A formula driven by headcount and years cannot see what is in your archive.

A firm offer comes after review. Polyshares puts it plainly: “We do not pay for a record we have not seen, so the review comes first.”

	Indicative offer	Firm offer
Based on	A company profile or calculator inputs	A sample, a rights check and the retained history
States	A range	A price, the records covered, permitted uses, exclusivity and the payment trigger
Commits the buyer to	Nothing	The written agreement, subject to its conditions

If a figure arrived before anyone saw a sample, treat it as indicative, whatever the email calls it. If the document names the systems, the date range, the permitted uses, any exclusivity and what triggers payment, you have something you can compare.

Why would two buyers offer different amounts for the same company?

Because each buyer is pricing its own demand, not a market rate. The largest single factor is whether a buyer has a customer for this kind of data right now. Most programs Data Licensing Report covers sit between you and the AI labs; Mercor, Appen and Troveo each describe licensing data onward. For example, a provider holding a current lab request for freight dispatch records can price a trucking company’s archive with some confidence. One without that request is estimating resale, and may price lower or pass.

Three other things move the number:

• Archive quality. Appen says it prices by “volume, the number and connectedness of systems, workflow complexity, data quality, and how rare the operational knowledge is”.
• Scope. Which systems, years and record types are in, as set out in a data inventory.
• Terms. An exclusive, perpetual license is a different product from a bounded, non-exclusive one. Miro Advisory lists “licensing scope, exclusivity, and whether future deliveries are possible” among what sets value. The payment model changes the headline too; see upfront payment or revenue share.

The plainest public example is a bankruptcy sale. For Spirit Airlines’ de-identified records, auctioned in August 2026, Bloomberg Law reported Google as the winning bidder at $10 million, with Mercor’s $7.5 million as the backup bid (Bloomberg Law). micro1 then offered $12.5 million for the same data after the bidding deadline had passed, as reported by Forbes and Fortune. As of late September 2026 the sale still needed the bankruptcy court’s approval (Moneywise). The three figures for the same records span $5 million.

What does a buyer check before making a firm offer?

Handshake AI’s estimator says any payout depends on “the outcome of our due diligence process”. That diligence has to answer five questions.

1. Can you license it? Troveo’s assessment asks whether the data is yours to license, with “Parts of it” covering data under client agreements or containing personal data (Troveo).
2. How far back does it go? Retained history is often shorter than the company’s age.
3. Does a sample match the description? See what a data sample is.
4. Can names, accounts and client details come out without breaking the record?
5. Who can export it, and with which tool?

When does the money arrive?

Most named programs pay on approval, delivery or acceptance rather than on signature. As of October 2026, they publish these triggers:

Program	What triggers payment	Published timing
Polyshares	Data approved, shared and anonymized	Net 30 to 60 (source)
Mercor	Wire on delivery of the anonymized dataset to the buyer	Typically 2 to 4 weeks after signing (source)
Corpus	Delivery confirmation	30 to 60 days from intake to wire (source); its get-paid page says 30 to 60 days after delivery (source)
Appen	Acceptance of the data, not submission	Not stated (source)
Nyne	Delivery and acceptance, per the agreement	Not stated (source)
Telegraph Lab	Payment milestones in the signed agreement	Not stated (source)
Troveo	Each sale of your data, as a revenue share	Not stated for business data (source)
Replay	Upfront cash plus a share of every license	Not stated (source)

Where acceptance is the trigger, the agreement should say who decides acceptance, against what standard and by when. Nyne says its acceptance criteria are “agreed up front”.

Why compare more than one offer?

Because one offer measures one buyer’s appetite on one day. Telegraph Lab has described one transaction to Data Licensing Report: for an online retailer, three buyers were approached and two made offers, and the owner received $300,000 within two weeks of the first conversation. It is a single deal, not an average, but one of the three buyers made no offer at all.

Compare on like terms: price, firm or indicative, which systems and years, exclusivity and its length, permitted uses, the payment trigger and, for a revenue share, the percentage and what it is a share of. A higher price for an exclusive, perpetual license can be worth less than a lower one that leaves you free to license again.

An agreement to deal with one buyer only, before it has priced anything, ends the comparison early; Polyshares, for one, states “No exclusivity while we talk”. The guide covers the rest of the process, and the offer benchmark lines up an offer you hold against these same fields.