Buy, not build
People sometimes ask why we acquire data platforms instead of building them. The honest answer is that the best ones cannot be built quickly, by us or by anyone.
A good data platform looks simple from the outside: sources go in, clean records come out. Inside, it is years of accumulated judgment. Which sources to trust, and when. Which of two conflicting records wins. The thousand edge cases that only appear at scale, each one paid for with a bad answer someone noticed. None of that compresses. You cannot hire it, and you cannot shortcut it. You can only accumulate it, one correction at a time.
Building from scratch means repaying all of that tuition, in the same currency, at the same rate. The industry has a long record of well funded teams discovering this the slow way.
So we buy. We look for four things:
- A working product, not a promising design
- Real customers whose work depends on the data being right
- A slice of the public record where staleness genuinely hurts
- A team that treats correctness as the product
In return we offer the one thing a fund cannot: permanence. We are not assembling a portfolio to exit. We operate what we acquire, and we intend to hold it for a very long time. A team that has spent years learning a hard corner of the record should not have to spend the next two wondering who owns them.
Buy, not build. It is less exciting to say, and much better to own.