As for Moneyball ideas for tenure, it would seem that behavior over the probationary period is the best predictor. If a candidate has done the research/publication as we expect in terms of contribution and quality, then ok. If not, no excuse seems to be exculpatory. The candidate won't become much better and may well decay rapidly.

More detail;

1. Web of Science won't work for book fields such as history, literature, some of anthropology, classics,..., so the analytics stuff won't work for a fraction of the university. Not even clear if citation data tells you much about these fields. Often the citations are implicit by the problems people take on.
2. My problem is to make better tenure decisions, in terms of stronger research careers. The only thing I can discern from a first rough reading of the literature is that the world is divided into producers and non-producers, and people evidence their status in the probationary six years. They do not change.

So one rule might be, if they don't produce, however you define it, do not tenure them.
3. The literature is concerned often with how institutional culture affects productivity. Mainly, like produces like.

4. There is a subtheme in much of this literature, that by production we want to enlarge it to included teaching service etc. Or maybe have a department in which some do one, some do another. Balance is a pervasive theme.

But my problem is how to increase research prominence and quality and quantity. I leave it to others to worry about life and family (actually, I don't think that really productive people are worse in their family lives than other faculty, and some of the research suggests that). Hence much of this literature is unhelpful. I am not a Stakhanovite, but I want to know how to make our university stronger. The answer, it would seem, is to insist on high performance during the probationary period, no excuses, and then figure out how to make the next period good to continue performance. (Put simply, Doubtful cases never turn out good.)

You might find better ways of hiring so that it is more likely someone will be research productive, and there is evidence that postdoc performance is important in the sciences, not so clear in other areas especially if no postdocs in these areas (!). And you could read the dissertation with some care, and decide whether it is first rate publishable work.