We mainly rely on review papers, meta-analysis and expert consensus, rather than single unpowered studies.

However, you don't need to have strong evidence behind a technique for it to be worth trying – if the upside is high enough or the costs are low enough, it could still be worth it.