Physical assets wear out. Patents have a 20-year clock and can be designed around. But proprietary data – for example, built through years of transactions, processes, and customer interactions – tends to become more valuable over time, not less. Most companies carry this asset without knowing its worth. Most lenders never ask. This piece examines why data IP is systematically under-recognised, what actually drives value, and what it takes to see it clearly.
A cleantech company secures a growth facility. The lender’s due diligence focuses on the patent portfolio – registered, documented, legally defensible. The data assets – years of proprietary power demand, pricing, environmental monitoring readings, and benchmarking datasets – go unexamined. Nobody models them. Nobody asks who might want them.
The data turned out to be four times more valuable than the patents.
This is the predictable consequence of a due diligence process designed around easily visible assets. Registered IP sits in a register. It has a filing date, a jurisdiction, an expiry. Data assets have none of these markers. They live in systems, processes, and pipelines that legal teams rarely access and financial models rarely reflect.
The lender secured against the visible fraction of the company’s intangible value. The rest sat there – appreciating.
A company is approached by a third party wanting to licence some of its proprietary datasets. The third party is serious. The interest is genuine. And the company has no idea what to charge.
There are no comparable transactions. No standard methodology. No internal view of floor or ceiling. The company built the dataset as a byproduct of its operations and never considered it something that could be priced and sold.
This is a good problem to have. Most companies never find out they have it – not because the interest isn’t there, but because they haven’t identified what they own clearly enough for anyone to ask the right question.
The pricing conversation, when it comes, happens on the buyer’s terms.
The obvious buyers of a proprietary dataset are the company’s own customers. They already know what the data can do; they just don’t own it. But the market is broader than that, and understanding it matters most when things go wrong.
In a sale or foreclosure, the buyer often won’t be the company’s customer. It will be whoever sees strategic value in owning that data outright – a competitor who wants to close an informational gap, a supplier who wants to understand the market it serves, an acquirer building a position in the sector. These buyers exist. Their appetite is often significant. And the price they’re willing to pay bears little relation to what the data cost to produce.
Realisability – the question of what an asset would actually fetch, in which scenarios, from which buyers – is the right frame for data IP. Most companies have never run that analysis.
Accounting rules make this worse. Internally generated intangible assets are expensed rather than capitalised under the standards. A company that builds a dataset over five years – investing in data infrastructure, data science, data cleaning – records no asset for any of it. The balance sheet is silent.
Buy a competitor with a comparable dataset and it may land on your balance sheet at fair value. Build it yourself and it doesn’t appear at all. The most data-rich companies are often the most misrepresented in their own accounts.
Data compounds this problem beyond what patents do. A patent portfolio at least appears on the register and at cost in the accounts. Proprietary datasets don’t appear anywhere unless someone has actively documented them. They grow in value while remaining invisible – to management, to lenders, to the market.
Seeing the data layer clearly is not a valuation exercise. It starts earlier than that. The first question is simply what exists: which datasets, built from what sources, structured in what ways, with what history of maintenance and governance. Most companies don’t have a comprehensive answer.
From there, the questions become more interesting. Who would want this? What would they pay in a going-concern context versus a distressed one? What is separable from the business and what isn’t? Is the data clean enough to transfer? Are there restrictions on how it can be used or sold?
These are not questions for the asset register. They require a different kind of inventory – one that treats data as an asset class in its own right rather than a feature of the product.
The companies that have done this work find themselves in a different conversation. With lenders who can see the full picture. With counterparties who want to licence something the company hadn’t thought to offer. And occasionally, with a much clearer view of what they’ve actually built.