1) The idea of the book. Is it original, new, different? Are there competing books on the same subject? How large is the audience for that subject, and is it an audience that your publishing house can reach effectively? And coupled with this is the expertise of the author to write about this subject.
2) Voice. Does the author have a distinctive, unique voice of his own, or is his writing bland, colorless, lacking pace and drama? We're talking here about the author's writing style, and how effective he is in presenting his ideas.
3) Facts. Has he researched the subject thoroughly and presented all the information needed to provide the most persuasive presentation of his thesis?
4) Organization. Has he found a way to put all this material together coherently and cohesively? One can imagine an author who is tireless in compiling research and information, but then doesn't know how to develop it all into a logical framework. Such manuscripts are usually hopeless as it would be an incredibly labor-intensive task for an editor to take such a manuscript and try to reshape it. But there are situations where a publisher might consider it necessary to do just that. In such a case, the publisher may call on free-lancers like the members of Consulting Editors Alliance.
5) Finally, has the author accomplished what he set out to do? I think of this as the combined effect of all the four other elements, and it's a crucial question to ask. An editor can't evaluate a manuscript on the basis of what he wants it to be. But of course if he doesn't agree with the author's objective, then the manuscript is probably not for him. On the other hand, if he does like the idea of the book, but feels that the manuscript has problems, then he must decide whether the author might be able to revise the manuscript satisfactorily or whether he thinks it could be salvaged with his own editing.