To build a great product you need a lot of rich feedback to iterate. The richness of this feedback is a function of its frequency and entropy (high entropy = less predictable). In a way, a product’s value is a combination of the effort to build it and the amount of feedback entropy that went into it. And what’s not intuitive is that it is the latter that’s dominant in that equation.
It is extremely difficult, borderline impossible, to design something great and new without information-rich feedback. A genius like Beethoven could compose symphonies after losing his hearing, but even that stemmed from his exceptional ability to simulate the music in his mind to effectively get the feedback (his own judgment) to iterate.
The issue is that it’s a lot harder to get early feedback for something truly innovative. High entropy feedback requires a lot of context, thinking about the problem space from first principles, and the ability to envision different paths for a product. It is hard to find.
Dogfooding (using your own product), is the best way to get over that problem. To me, it feels so fundamental that I would guess the amount of innovation in specific industries and problem spaces has a high variance for its dependency on the applicability of dogfooding.
It is so fundamental, that I would even claim you should design your company and product, to maximize dogfooding.