The committee members and the provost who read dossiers have enough experience collectively to figure out when you have not been forthright or have been deceptive. They'll check the h-index claims, they'll find out if NSF gives out grants in the field, they'll notice inconsistencies in the letters or the CV. Maybe not all will find them, but someone will. What amazes me is how often we find such tricks, deans or chairs or letter writers seem to feel they can get away with stuff. More to the point, maybe they are acting innocently, and that is even more frightening.

1. Letters should come from diverse institutions, and not have 3 of 6 or 10 letters from the same place, no matter how distinguished the place.
2. Letters that have identical paragraphs (this happens!) indicate there is an app for iPhone for writing promotion letters, or collusion or...
3. If a candidate for full professor has made slow but steady progress since tenure, eventually they ought be promoted. But at least be honest why. Strong contributions to teaching or administration may well be important at this level. But you probably won't be able to invoke all that until maybe a dozen years after tenure.
4. Lateral appointments should really be terrific. But: perhaps a senior member of the faculty needs someone, X, to continue his strong research program, and it must come with a tenure appointment or at least associate. For the strongest faculty, it may make sense to give them the chance to make the case. But it does not help to wax hyperbolic about X's independent research career, or X's grant getting, or make excuses for lacks. Rather tell the dean and provost and committee the real reason.
5. If someone who is non-tenure-track develops a portfolio of research and grants and publications that is quite impressive, it's fine to put them up for a tenure track appointment. If they are part of someone's research project, and have always been second banana, but that banana is wonderful, it might well be worth trying--with a frank account of why.
6. In some fields, grant support of a particular sort (R01's) is not about money about about peer judgment. There may be surrogates for that peer judgment, for cases where the research is paid for in other ways.
7. A faculty should be allowed one pecular case every decade. Not unqualified, but qualifed in an idiosyncratic way. Maybe they know something. But this case needs the weight of the whole faculty behind it.
8. I'm not sure how I feel about dean's having such a privilege separate from that of the faculty.
9. Sometimes retention cases arise for candidates who do not have strong records. The other institution sees virtues in the candidate that are not perceived by your own, or perhaps they have need for particular strengths. If the dossier is not convincing to you, do not vote yes--let the person go elsewhere, give them a goodbye party that is heartfelt, and everyone is better off.

More generally, if you have peculiarties in a dossier, take charge of them. Excuses won't work, but reasons may. Never assume that problems will not be noticed, never assume that peculairities won't be noticed.

Finally, your credibility is on the line. If you send up dossiers that are fishy, you begin to smell like old cod, and then in cases where you wish to really be able to exert your one free pass, so to speak, no one will give that benefit of the doubt.