The case for competing offers
One buyer names one number. Several buyers reveal what the records are worth. A widely reported 2026 bankruptcy auction shows the gap, and the same logic applies to a company that is not for sale.
The best public illustration of why competing offers matter did not come from a data company. It came from a bankruptcy court.
In 2026 the business press reported on the sale of the data assets of the former Spirit Airlines: by the reports, roughly 100 million emails, 500 million chat messages, and 30 million lines of code. Google was reported to have opened the auction at $5 million. After several rounds, the airline reportedly selected Google's $10 million bid, with the AI data company Mercor named as the alternate at $7.5 million. Days later, another AI data company, micro1, was reported to have submitted a late offer of $12.5 million, with the outcome left to the court. Reported figures, reported sequence, and unsettled when this was written. But the shape is the thing to notice: the same records, described once, moved from $5 million to a reported $12.5 million because more than one buyer wanted them on the same deadline.
One buyer prices for the buyer
When a single buyer approaches a company about its records, the number in the email is not a valuation. It is the amount that buyer has decided is worth paying for this category of data this quarter, discounted for the chance that the owner will say yes without asking anyone else.
That is not dishonest. It is how any buyer behaves when the seller has no reference price. The problem sits on the seller's side: most owners have never licensed data before, do not know who else buys it, and have no way to tell whether the number is generous or a third of what the records would draw elsewhere.
A single inbound offer therefore carries two unknowns. Is the price right, and are the terms right. Without a second buyer, neither can be tested.
What competition changes
Price is the obvious part, and the reported auction shows how far it can move. The less obvious part is terms, which in data licensing often matter more than the headline figure.
Exclusivity. A buyer who is the only bidder asks for exclusive rights by default, because exclusivity costs nothing to request. A buyer competing with two others will often accept non-exclusive terms, which means the same records can be licensed again to a buyer in a different field.
Field of use. One buyer wants the records to train a customer-service model; another wants them for evaluation only. Splitting the use between them is possible only when both are at the table.
Term and deletion. The length of the license, and the date by which the supplied copy is deleted, are negotiable when there is a second buyer and fixed when there is not.
Payment timing. Competing buyers tend to shorten their payment terms. A sole buyer tends to lengthen them.
Each of those moves the real value of the deal to the owner, and none of them appears in a one-line inbound offer.
Why most owners do not run one
Three reasons come up every time.
They do not know the buyers. The companies that license operational data are not household names. There are frontier labs, specialist data labs, model builders, and enterprise AI teams, each with its own programs, and the list changes every quarter.
They do not have the time. Running a process means describing the data once, approaching several buyers, handling questions, and holding everyone to one deadline. A busy owner who gets one offer on a Monday is tempted to decide by Friday.
They are worried about exposure. The instinct is that talking to five buyers means five companies looking at the data. It does not, if the process is run correctly.
Competing without exposing anything
A competitive process is run from a manifest, the one-page inventory that describes the records without containing any. Every buyer that fits gets the same page, on the same day, with the same deadline. They price from that page. None of them sees a record.
Only after a first offer comes back, and only for the buyers you approve, does a small de-identified sample follow, under a signed agreement that governs its use. The dataset itself stays in your systems until a license is signed and then moves directly to the buyer you chose. At no point does it pass through the agent running the process.
Done this way, competition costs the owner an afternoon to approve the manifest and a few decisions along the way. The buyers do the work of competing.
The deadline that is really a red flag
One tactic is worth naming because it is designed to prevent exactly this. An offer that expires in 48 hours, or that must be accepted before anyone else is consulted, is not an offer with a strong price behind it. A buyer confident in their number does not need to stop the seller from checking it. Treat the deadline as information, and take the two weeks anyway.
What we do with this
Our process starts by pricing the manifest against what vetted buyers are paying for comparable records right now, before any buyer sees your name. That produces a first offer, an indicative and buyer-benchmarked range, within 24 to 48 hours of a complete manifest. Then the qualified set goes to more than one buyer at once, with one deadline, and binding terms follow sample review. We think the right number of offers is three, not one, and the reported auction above is the plainest public case for why.
If a buyer has already approached you, the manifest is still the first step. It tells you what the number should have been.
Questions worth asking
Will shopping an offer to other buyers offend the buyer who approached me?
A serious buyer expects it. Data licensing teams run competitive processes all the time from the other side, and a seller who asks for two weeks to benchmark is not a surprise. A buyer who reacts to that request by shortening the deadline or withdrawing is telling you something about the offer. The polite version is simple: you are interested, you want to understand the market value, and you will come back with a decision on a date you name.
Does a competitive process expose my data to more companies?
Not if it is run from a manifest. Buyers price from a one-page inventory that contains no records, and only the buyers you approve ever see a small de-identified sample, after a signed agreement covering its use. Competition happens on the manifest and on terms, not on the dataset. Your data never passes through the agent and moves only to the buyer you sign with, after signing.
What if only one buyer is interested?
Then you still want to know the benchmark before you answer. Pricing the manifest against what vetted buyers are currently paying for comparable records tells you whether the one offer is fair, low, or unusually high, and gives you a number to negotiate from. A single bidder is not a reason to accept the first number. It is a reason to make sure the first number was priced from the market rather than from the buyer's budget.
Measured from a complete data manifest. A first offer is a process target, not a guarantee; buyers decide eligibility, price, and terms. Indicative value ranges mentioned on this site are illustrative, not quotes. Nothing here is legal advice; have your own counsel review any agreement before data moves.