[manmal]

Depending on where you are coming from (sceptic, or a more "relaxed" view of unscientific proceeds), you can try several things. Over at ubiome.com you can order a microbiome test and see what strains you are missing or deficient, and then go and take exactly those strains. They might be hard or impossible to come by, but this might be the more targeted approach.

What I did was just throwing lots of different (valid!) bacterial strains in, and see what comes of it (ubiome wasn't available back then, and regular lab tests are quite expensive). Kefir did the most for me, and it has different L.bacillus strains and some yeasts in it (http://ntur.lib.ntu.edu.tw/bitstream/246246/177589/1/04.pdf). Kefir is reported to kill off or starve strains that don't belong in its ecosystem, but it's quite possible that this is not true or that there is point beyond which the L.bacilli will die off and the unbeneficial strains take over. Smell and look are reportedly good indicators for that, but I didn't find any studies for that. One would have to send it into a lab to make sure.