Empirical studies on revision quality show that the added value of revision is often low, or even nonexistent. Revisers tend to miss many problems, make unnecessary changes and sometimes introduce errors into the draft translation. This alarming but scientifically-grounded picture calls for explanations and solutions. The aim of this article is to understand the situation and come up with potential solutions by drawing upon a large interview study that was carried out in various settings, mostly in translation teams. One factor is that revision is difficult for professional translators, as shown through nine difficulties cited by participants in the interviews, including limiting the number of changes and staying focused on the task. The poor results observed in empirical studies may also stem from poor training or the lack of supervision of revisers. As for solutions, in addition to promoting training and supervision, this article also advocates for advisory revision and oral revision, instead of corrective, written revision.