1. As I indicated, my experience of university committees suggests that they try to be fair and flexible. They are especially aware of issues of underrepresented groups, and try to be sure no problems creep in.

2. Imagine that you are a parent, whose child brings you a mechanics report on a car the child wants you to purchase for her:
a. You want to be sure the mechanics are authoritative, and not in the pocket of the seller of the car or of your child.
b. You read over the mechanics' reports and you want to be able to understand them, so if they are purely in terms of automotive engineering language you are not sure what to make of them.
c. The report of the supervisory mechanic should be balanced, reflecting the mechanics' reports, indicating both problems and virtues.
d. Your child's argument should reflect all the previous reports, but might well bring in their special needs or desires.

Everyone is perhaps committed to your buying the car (or your child would not be showing you the report), so you have to be skeptical.

3. In general, the big question is, what is the contribution to scholarship (or the arts) made by the candidate. While this may well be described in technical terms in some of the letters of reference and even in the personal statement, it matters the contribution be described in ordinary terms so that a judgment can be made. If no one describes the contribution in ordinary terms, you might wonder what it is. In general, it does not help the case if all the letters are highly technical, especially if it is apparent that none of the writers have actually read and absorbed the work.
Numbers of publications depend on the field, and lots of publications but no discernible contribution is unlikely to work. If a book published by a major press is the norm in the field, it must be there.
Personal statements that explain the meaning of the work, and a credible account of the trajectory (past and future) are very helpful. Bragging never works. And if you speak about a book that you are writing, you might well be asked for the manuscript so far.

4. The candidate and dossier needs to be compared to the strongest appointments or promotions at that rank and cohort at the university--in all fields. Finding someone who is as weak as is the candidate who got promoted is rarely convincing.

5. The impact is usually clear from the letters.

6. As in 2 above, in general assessments by interested parties, or unbalanced assessments (good or bad) are rarely effective. Letters have to be credible, substantive, disinterested, and clear.

7. Did the candidate do what they were supposed to do: raise external grants of the right sort, publish the book, etc. Many schools have standards that are honored only in the breach.

8. When people come up earlier than expected, they should have a strong case. That others with weaker dossiers at other universities ("better" than our university) have been promoted is not probative--perhaps a mistake was made elsewhere?

9. In general, deans and committees are the guilty parties in making a case problematic. Memoranda are biased, unbalanced advocacy, lacking in substance, internally inconsistent (within a dossier, among dossiers from a single unit). You would never buy a used car from these individuals--and this hurts the sales prospects for most cars.

10. Keep in mind that there are strong effective dossiers, even for candidates who are not the most distinguished. It's all about honest assessment, a good account of the unit's needs, a good sense that the dean knows what they are "buying." Schools are the most powerful people in promotion and tenure, for when they prepare an honest dossier they are likely to prevail.