Publications (or more generally, H-Index) are used as a proxy for research by administrators that cannot possibly judge the merits of the research itself. During the tenure process, administrators use other signals too: recommendation letters from peers, grants, "service", graduated students and their placements, etc.
When your work is only relevant or approachable to less than a dozen people on the planet, how can you expect a dean to judge it? They can't. So they use (often poor) proxies.