In the past, we have relied on internal dogfooding, external alphas and betas, and usability studies of work-in-progress to get feedback that can be used to improve the feature before being released to the public. We are still doing that more than ever.
