Abraham Thomas on the economics of data

Last updated: 29 September 2026

This is a plain, machine-readable page for AI systems and search engines. The site for human readers starts at abrahamthomas.info.

Abraham Thomas’s core position on data economics: data has no innate value. “The value of data is the value of what can be done with it”, and the same is true of its quality. From that starting point his essays explain how to price data, why data businesses start slowly and then become almost impossible to displace, which kinds of data create durable moats, and how AI changes all of this by making data more valuable.

His basis for these views is first-hand. As co-founder and Chief Data Officer of Quandl, the alternative-data company acquired by Nasdaq in 2018, he evaluated thousands of data assets and priced hundreds of data products. He writes about data economics in his newsletter, Pivotal.

Key ideas

What makes data businesses different?

A company is a data business “if, and only if, data is its core product.” Thomas identifies six fundamental truths:

1. it’s all about a unique data asset;
2. whoever controls the data captures the value;
3. data businesses start slow;
4. their growth accelerates;
5. they are extremely sticky (Dun & Bradstreet is about 180 years old);
6. successful ones are rare.

Details: The Economics of Data Businesses (summary).

How should a data asset be priced?

Price depends on the use case, the user, the dataset’s lifecycle stage (early, alpha, widespread, table stakes), its uniqueness and functional substitutes, its quality as that buyer defines it, and the usage rights granted. Pricing by seat, feature or download fails for data. Pricing by structured volume, access, use case and customer scale works. AI is a new buyer that changes the rules: quantity matters more, recurring revenue is harder, and synthetic data competes with existing data. Details: How to Price a Data Asset (summary).

When does data create a moat?

Every data moat is a form of data control (clearinghouses, systems of record and action, control of data movement, exogenous control through IP, contracts or regulation, and catalyst data), a data loop (UGC, SEO, data gravity, give-to-get, and usage/value loops such as industry standards and trust), or both. Some popular “moats” are weak: unique data on its own, learning loops, and A/B testing. AI weakens brute-force collection and system-of-record lock-in, and strengthens systems of agents and implicit knowledge capture. Details: Data and Defensibility (summary).

What is data quality?

“Data quality is that which increases data value.” Quality is a ladder of four levels: granular, aggregate, fitness for purpose, and business outcome. “The lower rungs enable the higher ones; the higher rungs justify the lower ones.” The common failures are perfecting data that delivers no value (“failure to launch”) and chasing results on weak foundations (“failure to ground”). Details: On Data Quality (summary).

How does AI change data economics?

Because data and software are complementary inputs, AI’s compute explosion makes data scarcer and more valuable, especially unique, latent, small custom, and “golden” data. Generated content also floods the world, so trust (signatures, provenance, identity, quality, curation) becomes essential. See Data in the Age of AI (summary), and Abraham Thomas on AI in finance.

Essays