A single offer tells you what one buyer wants to pay, on terms it wrote. Two or three offers, made against the same manifest in the same window, tell you what your data is worth and which terms are negotiable.
Last checked: 7 October 2026. Buyer facts are quoted from their own pages; market figures are attributed.
Four things you cannot know with a single buyer at the table.
Published ranges are wide. micro1 lists "$100k+ qualified" up to "$1M+ highly unique"; Mode lists "$100K-$5M"; Grepped lists "$20K-$5M." Those are "up to" ranges, not averages. Where you land inside them is exactly what a second offer reveals.
The first draft of any agreement reflects the drafter's preferences on exclusivity, scope of use, liability and payment timing. Without an alternative, every change you ask for is a favor.
If a buyer asks for exclusive rights, a non-exclusive offer from someone else shows what that exclusivity is costing you. See exclusivity and resale rights.
One program may care most about your support tickets, another about code history or finance workflows. A single offer prices the whole dataset through one buyer's priorities.
The one public example of competition is a bankruptcy auction. According to a September 13, 2026 report by Stretto on chapter11cases.com, Google opened the Spirit Airlines data auction at $5 million and was named winning bidder at $10 million, with another bidder as alternate at $7.5 million. micro1 then made a $12.5 million rival offer. The same report quotes the sellers' declaration that non-economic factors, such as "a documented and acceptable deidentification process," may be "outcome-determinative." Read the full Spirit Airlines case file; the lesson for you is that price moved with competition, and terms decided between close bids.
Practitioners say: never send a full dataset before price. Share a manifest and samples, and get more than one offer.
The same document goes to every buyer, so offers are comparable. It describes the data; it contains none of it.
A few representative threads, tickets or records with names and identifiers removed. Enough to judge quality, not enough to train on.
Check each program's published eligibility first. micro1 lists 30+ employees; Mode lists 20+ full-time US office employees (10+ for accounting firms, 6+ for law firms); Grepped lists any vertical. Starting together keeps the timelines aligned.
Look for any exclusive negotiation period or no-shop clause. If one is requested before you have an offer, ask how long it lasts and what you get in return.
Set your own decision date and tell each buyer when you will decide. A deadline you state is more useful than one you are handed.
Use the table below. Then go back to your preferred buyer with specific asks drawn from the other offers.
One to two pages. Enough for a buyer to value it; nothing that exposes a person or a client.
| Field | What to write | Why buyers ask |
|---|---|---|
| Systems | e.g. Slack, Jira, Zendesk, Salesforce, GitHub | Buyers list these sources by name and value connected systems |
| Date range | First and last month of records per system | Several years of history is a published criterion (Mode) |
| Volume | Approximate messages, tickets, documents, repositories | Sets the scale of the deal |
| People | Number of users and teams; primary language | micro1 lists "primarily English" teams |
| Workflows shown | Which processes the records capture end to end | micro1 lists SOPs, project histories and "decision-making patterns" |
| Exclusions | DMs, HR, client-confidential matters, customer PII, secrets | Shows the buyer what it will not receive, up front |
| Constraints | Client contracts, consents, sector rules that apply | Avoids a late surprise that kills the deal |
Buyer criteria as published on micro1.ai/data-partnerships and data.mode.inc, checked 7 October 2026.
The sample that goes with it. A good sample is small, representative and safe. Pick a handful of complete examples rather than many fragments: one support ticket from first message to resolution, one project thread from kickoff to delivery, one month of a recurring process such as a reconciliation. Complete examples show buyers the connected history they say they value; fragments do not.
Remove or replace names, email addresses, phone numbers, client names and anything a person could be identified by. Use consistent placeholders (the same person is always "Employee 4") so the buyer can still follow who did what. Leave out anything from the exclusion list, even if it would make the sample more impressive. If a buyer asks for more, give more of the same kind, still redacted, rather than a raw export.
Keep a log of which buyer received which sample and when. If you later sign with one buyer, you will want to ask the others to delete what they received, and the log tells you exactly what to ask for.
When you have chosen, tell the other buyers promptly and politely, and ask them in writing to delete the samples under the NDA. Markets change, and a buyer you turned down this year may be the right one for a later dataset.
Questions every seller should check in each offer. These are generic, not claims about any named buyer.
| Term | Ask of every offer | Why it changes the value |
|---|---|---|
| Payment structure | One-off or recurring? Milestones? What are the acceptance criteria? | Money paid on acceptance depends on a standard you should see in writing |
| Exclusivity | None, time-limited or perpetual? Does it cover similar data you create later? | Perpetual exclusivity removes every future buyer |
| Scope of use | Training only? Evaluation? Which downstream buyers? | Broader use is worth more to the buyer, so it should cost more |
| Resale | Can the buyer resell or sublicense the copy? | Resale multiplies where your data ends up |
| De-identification | Who does it, to what standard, and can you audit it? | Errors here are the main source of risk after the sale |
| Liability | Are indemnities capped? Do warranties expire? | An uncapped indemnity can exceed the price you were paid |
| Consent representations | What do you promise about employees, customers and clients? | Promises you cannot verify become your liability |
| Deletion | What happens to originals and to the copy, and when? | Sets how long your data exists outside your control |
| Termination | Which clauses survive if the deal ends? | Surviving obligations outlast the payments |
Go deeper: indemnities and warranties in data deals and questions to ask a data buyer.
Two fictional offers for the same 120-person company's support and project records.
Offer A is a third higher, but it removes every future buyer, puts open-ended risk on the seller and pays only against criteria nobody has written. With B in hand, the seller can go back to A and ask for time-limited exclusivity, a liability cap and written acceptance criteria. Without B, there is nothing to point to.
Data deals move through NDA, buyer review, agreement, export, de-identification and acceptance. Practitioners cite 60 to 90 days to close. A buyer that needs an answer within days is asking you to skip the comparison, and you are entitled to ask why.
Use deadlines in your favor instead. Start all conversations in the same week, give every buyer the same decision date, and make the date real. If one buyer moves faster, you can tell the others when you expect to decide, without sharing the details of anyone's offer. Treat terms you received under NDA as confidential.
The method is the same. What changes is how many buyers you can realistically reach and what you have to offer each one. Minimums quoted from each program's page, checked 7 October 2026.
You meet Mode's published 20+ and micro1's 30+ minimums, and Grepped lists any vertical. That is already three programs to approach in the same week. Your manifest will be short, so make it precise: exact systems, years and the workflows they show. One owner, usually the founder or finance lead, can run the process in a few hours a week.
Several departments means several possible slices: support, sales, finance, engineering. Different buyers may bid on different slices, so ask each whether it wants the whole set or part of it. Comparing a whole-set offer with two partial offers is a real option at this size, as long as no exclusivity clause blocks it.
You can add direct lab intake to the data company programs, which widens the field. You will also have procurement, legal and privacy teams who need to review every NDA. Run it like any vendor selection: one manifest, one timetable, one comparison table, one decision meeting.
None of these depends on the buyer. All of them are in the seller's hands.
The first offer expires before the second arrives. Start everyone in the same week, with the same documents.
If each buyer saw a different scope, the offers cannot be compared. Use one manifest and note any slice a buyer asks about separately.
An exclusive negotiation period before any number is on the table ends the comparison before it starts. Ask for an offer first.
Exclusivity, liability, payment timing and scope of use can outweigh a price gap, as the worked example above shows.
A sample large enough to train on is not a sample. Keep it small, redacted and representative.
You can say you have other offers and when you will decide. Do not forward another buyer's terms if an NDA covers them.
Owners, the board and counsel should see the comparison table before the decision date, not after a verbal yes.
A short memo of the offers and the reasons protects you later, when staff, clients or investors ask how the decision was made.
Applications go through each program's own form. Once a buyer responds, these are the points to cover in writing.
Company size, country, industry and primary business language. These map to the published eligibility rules, so the buyer can confirm fit quickly.
The manifest, attached. Systems, years, volumes, workflows and exclusions.
That you are speaking with more than one buyer, the date you will decide, and that you will compare offers on terms as well as price.
A written offer with a validity date, a draft of the key terms (scope, exclusivity, de-identification, deletion, liability, payment), and the steps from signature to payment.
Send the full dataset before price and terms are agreed. Saying so early saves both sides time.
Yes. With one offer you have no benchmark for price or terms. Practitioners advise sharing a manifest and samples, never the full dataset, and getting more than one offer. In the one public auction, Spirit Airlines, the reported price rose from a $5 million opening bid to $10 million.
A short inventory of what you could license: each system, its date range, approximate volume, number of users, what the records show, and what you will exclude. It lets buyers judge value without seeing the data, so you can show it to several of them.
Each program reviews eligibility separately, so you can ask more than one whether your company fits. Before you sign anything, read every NDA and term sheet: some agreements include an exclusive negotiation period, which would limit you for its duration.
Put them side by side on the same rows: total price and when it is paid, exclusivity, scope of use, resale, de-identification and audit, deletion, liability cap and how long warranties survive. A higher headline price with uncapped liability or perpetual exclusivity can be worth less.
There is no standard. Ask each buyer for a written offer with a validity date, and set your own decision date so all offers arrive in the same window. Practitioners cite 60 to 90 days to close a deal, so plan for weeks, not days.
Yes, in general terms. Saying that you are speaking with more than one buyer and will decide by a given date is normal and helps keep timelines aligned. Do not share another buyer's specific terms if they were given to you under an NDA.
Then you still have learned something: whether other programs found you eligible, and why not. Use the published ranges and the comparison table to test the single offer on terms as well as price, ask for changes where the terms are weak, and do not let a deadline push you into signing before your counsel has read it.
Possibly, if no exclusivity clause prevents it. Some buyers may want only part of what you have, such as support tickets or code history. Ask each buyer which slices it values, and check every offer for exclusivity terms that would cover similar data sold to someone else.
Check which programs your company fits by their published rules, then apply to more than one so the offers arrive together.
Independent site. Some links are referral links: if your company signs with a buyer through them, the buyer may pay us a fee. You are not charged, and we never see your data.