Skip to content
CONCEPTLAST REVIEWED 2026-09SOURCED FROM 2 SESSIONS, APR–MAY 2026

A data moat is defensibility that comes from data a competitor cannot cheaply reproduce. The two repeatable sources named in the sessions: collecting what nobody publishes, and decades of domain fluency about what is signal and what is noise in a niche. History cannot be re-collected, which is what makes a long-running dataset unassailable.

It is the one ingredient a small company can own.

It’s much better than… trap them, then please them. Right? Trap them in the correct way.

Managing partner, deep tech seed fundsession, Apr 2026, on products users can’t leave

The flywheel mechanics from the practitioner roundtable: make feedback pay instantly, because users label your data when doing so immediately improves their own experience; interview your delivery and support people about what context the dataset lacks; back-test creative sources against outcomes; and start small, since a small dataset surfaces the strongest signals while granular personalization is what needs scale. When you pitch it, skip the nuance: largest dataset in X, trains the best models, customers stay because the product keeps getting better.

  • Commodity data as moat.
  • Waiting for scale.
  • Relying on goodwill.
  • Over-explaining to investors.