> identifying a subset of the 1000s of features used that a cover a large enough part of the market Deceptively hard. The only way to make this happen is usin

> identifying a subset of the 1000s of features used that a cover a large enough part of the market Deceptively hard. The only way to make this happen is using telemetry, and users don't like that, understandably. My belief is that Excel and Word are so successful because EVERYONE uses a different subset of those thousands of features. A one-lawyer office will differ from a dozen-person firm, a big general contractor will use different features from a small one, etc. You can't survey users about those features either. Users don't know what your surveys mean half the time. They will answer according to how they think you want them to answer instead of what they really do. Or they will be satisfied but name features they would never use because they heard about them at a conference once. They use different terminology from you, so signal gets lost either way. Ad nauseam.