What data monetisation used to mean
For most of the last decade, data monetisation meant getting more out of the data you already had. MIT researchers Barbara Wixom, Cynthia Beath and Leslie Owens describe three routes: improving operations with data, wrapping data into products (recommendations, delivery estimates), and selling data-based offerings such as insights and information services. Their research found 82% of data monetisation returns came from improving and wrapping, and only 18% from selling (MIT Sloan). In practice, for an Australian SME, the old version usually meant a better dashboard, a reporting add-on for customers, or, at the shady end, selling a customer list.
That last route is mostly closed, and should be. Customer names, emails, addresses and phone numbers are personal information under the Privacy Act 1988 (OAIC). A customer list is the one dataset most owners think of first, and the one that carries the most obligation and the least lasting value.
What data monetisation means in 2026
The new meaning is more literal: a business licenses a prepared copy of its operational record to an AI lab, or to a company that builds training environments for one. The first wave of this market was content. Reddit licensed user discussion to Google for a reported ~US$60M a year, and News Corp signed with OpenAI for a reported US$250M+ over five years (what data actually sells for).
In 2026 the same buyers moved to company records. When Spirit Airlines was liquidated, Google agreed to pay US$10 million for its corporate records and data, Mercor sat behind it as backup bidder at US$7.5 million, and micro1 then offered US$12.5 million. Fortune reported the archive held about 100 million emails, 500 million Microsoft Teams items, code and operational records (Fortune, 14 September 2026). On 9 October 2026 micro1 committed US$1 billion over the following 12 months to acquiring and licensing enterprise operational data, saying it wants records that reflect "incomplete information, competing priorities and exceptions that require judgment" (micro1).
Why buyers pay for operational records, not customer data
AI labs are training models and agents to do real work inside real businesses. What those models lack is the record of how work actually happens: a job comes in, someone quotes it, a decision gets made with partial information, and an outcome follows. The public internet has plenty of opinions about how to run a business and very little of the evidence. A ticket history with resolutions, quotes won and lost, inspections and the maintenance that followed, internal email and chat where decisions were argued out: that is the evidence.
Customer data teaches a model very little about that. A list of who bought what is not a record of judgment. It is also where the legal risk sits. The objections to the Spirit sale came from unions and consumer advocates arguing that personal information of former employees and customers was in the data set (Fortune). That is why Wildcatt wants operational data, meaning how decisions were made and what outcomes they led to, and does not want customer lists, contact details or purchase histories. Where people appear in the working record, they are removed before any buyer sees it. The detail is in our guide to de-identified data.
The routes to monetise data
Licence or sale. A licence grants a buyer defined rights over a prepared, de-identified copy: permitted use, term and territory, written down. The business keeps ownership of the original and keeps operating exactly as before. An outright sale transfers the records themselves, which is what happens in a liquidation like Spirit's, where there is no going concern left to keep them. For a trading business, a licence is almost always the right structure.
One buyer or several. A non-exclusive licence lets the same prepared copy go to more than one buyer, each approved by you. An exclusive licence to one buyer closes off the others for the term, so it has to be priced as such. Which is better depends on the dataset and on who is buying at the time; there is no rule that holds for every business.
What drives the value when you monetise data
- Years of history. Long, unbroken records show how decisions played out across cycles. Gaps and abandoned systems discount the price.
- Linked systems. Email plus job records plus quotes plus accounts about the same events is a connected record of how work happened. One system on its own is closer to a commodity.
- Outcomes recorded. The job was completed or abandoned, the quote won or lost, the fault fixed or recurring. Records without outcomes show activity; records with outcomes show judgment.
For a sense of scale, micro1 says qualified partnerships earn more than US$100,000, large datasets or ongoing work across several teams more than US$500,000, and highly unique proprietary data more than US$1 million (The Next Web). Those are one buyer's published ranges in US dollars, not a quote for any Australian business. The five drivers are set out in full in what your data is actually worth.
What it takes
The distance between a business's systems and a signed licence is four problems: clean rights to what you hold, privacy handled properly, the data structured to a buyer's specification, and access to the buyers. Rights come first, because material you hold for clients or under third-party licence was never yours to license. Privacy comes next: in Wildcatt's process, more than 60 identifier types are irreversibly removed with a full audit trail before any buyer sees anything, and extraction is read-only, so nothing in your systems changes.
The first practical step is knowing what you hold. The data inventory guide shows where the record count lives in each system Australian businesses use, without exporting anything. If the record is worth more than it costs you to keep, it belongs in your thinking as a business asset, not a storage bill. There is nothing to pay to find out.
Data monetisation used to mean squeezing value out of customer data. The version that pays in 2026 is licensing the record of how your business makes decisions, with every person in it removed.
Related guides: Data as an asset | De-identified data | Selling company data in Australia | Instant range
Sources: MIT Sloan, Data Is Everybody's Business; OAIC, What is personal information; Fortune, 14 September 2026; micro1, 9 October 2026; The Next Web, 9 October 2026; content deal values as listed on what data sells for. General information, not financial or legal advice.