If you want fewer syllables per word, you'll want a larger number of possible syllables. (For a metaphor, think about how many letters vs how many kana vs how many kanji you need to represent a particular Japanese word. The more possible glyphs/syllables you have, the fewer of them you need to convey the same amount of information.)
Some good ways to do this:
- Allow lots of different coda consonants.
- Have lots of different vowels.
- Allow clusters of two consonants in onsets or codas, instead of just one.
For example, if you start out allowing only CV syllables, and then you decide to add long vowels, that doubles the number of possible syllables. If you allow CVn instead of just CV, that doubles it again. If you allow sC instead of just C, that's another doubling…
This is why English has over ten times as many common syllables as Japanese (Oh's corpus analysis gives 6,949 vs 643 in the 20k most frequent words), and thus why English words consist of fewer syllables than Japanese ones. We have a whole lot of vowels, many possible codas, and very elaborate clusters in both onsets and codas (consider "strengths"). Japanese only allows two possible coda consonants (N and Q) and the only valid onset cluster is Cj.