The fifteen questions we hear most, answered the way we would across a table.
Yes, when done properly. A data sale is an ordinary commercial licence over property your company owns, executed under written terms with warranties. Australia's licence-first copyright setting means this is exactly how the market is supposed to work here. What makes it lawful in practice is the order of operations: rights confirmed first, personal information irreversibly removed second, and only the prepared, de-identified dataset ever reaching a buyer. See the rights and privacy guide.
They are protected by design, not by promise. Before anything leaves your environment, we run irreversible de-identification across 60+ categories of identifiers, with every removal logged in an audit ledger. Customer-owned content is excluded entirely at extraction. Buyers additionally sign non-reidentification obligations. What a buyer receives is the shape of how your business operates, with the people removed.
AI labs and AI companies building models that need to understand how real organisations work, plus, for some datasets, industry buyers who license benchmarks and market signals. This is not theoretical: Reddit, News Corp, Shutterstock and major academic publishers have all disclosed deals. See what data actually sells for.
It depends on five drivers: years of continuous history, how many systems you run, whether your rights are clean, the quality of the activity, and how rare your kind of data is in existing training sets. Honest answers range from "not much yet" to sums material against annual profit. The assessment exists to replace guessing with a defensible range. See the valuation guide.
It is short, fixed-scope and priced as such: agreed up front, no day rates, no open-ended engagement. You get an inventory of what you hold, a valuation range against the market's pricing drivers, and a plain recommendation, including "leave it in the ground" if that is the truth. Beyond the assessment, we price on outcomes: we earn when your data does.
Read-only access to the systems in scope, granted by you, revocable by you. Nothing in your systems is modified, moved or deleted; extraction is a copy operation with an audit trail. Your team's involvement is roughly an hour of admin to grant access. See how your data is handled.
The assessment runs in weeks, not months. If a sale proceeds, preparation (extraction, de-identification, structuring) runs in days per system thanks to the technology we built for it, and the overall timeline is then set by the buyer's review process. By hand the same preparation takes months; that is precisely the problem the technology removes.
No, it is normal, and it is often good news. Value concentrates in cross-system history: the same events visible in chat, tickets, code and CRM. "Messy but continuous and connected" beats "tidy but thin" every time. Gaps and system migrations get assessed honestly; they discount value rather than destroy it.
No. Buyers are not reading your dataset for market intelligence; they are training models on de-identified operational patterns aggregated with many other sources. Your customer lists, prices and identities never appear, contractual use is limited to the licensed purpose, and your business keeps every byte it had. Nothing transfers that a competitor could act on.
Usually the opposite: a wind-down is often the cleanest moment to realise the data asset, and it may be the estate's most recoverable value. The one genuine risk is time: SaaS accounts lapse and get purged, so access needs preserving early. We work directly with appointees and advisors, with documentation prepared for the file. See the practitioner's guide.
We find the buyer and run the deal on your side of the table. We are not the buyer, so we are not pricing your record against our own margin, and we are not a marketplace, so nothing is listed anywhere public. Buyers are AI labs and AI companies we approach directly with a de-identified profile of what you hold. We are paid on outcomes: when your data earns, we earn, and the assessment is the only fixed fee in the process. The trade-off is honest: a direct buyer can quote faster because it is quoting its own money; a deal partner gets you the market's price rather than one buyer's.
Always a licence, never a transfer. A sale prices the record once, at the size it is today, and takes ownership with it. A licence grants defined rights over a prepared, de-identified copy for a defined period, and your business keeps every byte and every right. Three terms set the price: exclusivity (an exclusive licence pays more and closes the record to other buyers for the term; a non-exclusive one can be granted again as the record grows), term (perpetual over a fixed cut of history, or fixed years with renewal), and permitted use (training, evaluation, research, territory, no re-identification, no resale). All three are your call, with our recommendation beside each. See how a deal is structured.
No. The intake asks which systems you run, roughly how much sits in each and how many years they cover. Estimates from each system's admin console are enough; this guide shows where the counts live, system by system. Nothing leaves your environment at intake or during the screen. Read-only access comes later, for the assessment, and extraction only after a licence is signed.
No, it is the normal case. Most businesses hold two kinds of material: their own operating record, which they own, and material held for clients under master agreements or terms of use, which belongs to the client. Only the first is licensed. Client-owned content is excluded at extraction, and where a customer agreement permits aggregated or de-identified use, that is the clause most deals run through. The first question in any screen is "which of this is actually yours", and it is the question that stalls most owners who try to do this alone.
You know the counterparty before you sign, because you sign the licence. Until then, buyers see a de-identified profile of the record: category, years, systems, scale. Nothing is listed publicly, and the dataset itself carries no names, contact details or identifiers for staff, customers or counterparties, so there is nobody in it for a buyer to contact. The licence adds the contractual lock: no re-identification, no resale, use limited to the permitted purpose.