Original measurement.

Studies I run myself, on markets I work in. Each one publishes the sampling frame, the method, the confidence intervals, what the data cannot support, and an aggregate dataset anyone can re-analyse. Where a finding contradicts what I sell, it still gets published.