Don't know if this is too late to get a reply from the staff here, but
I noticed you mentioned the levels are maxed at -3db, which sounds reasonable.
But I was wondering if you'd put any audio compression on them, you didn't mention it.
I've not actually noticed a particular problem myself, but if other people say they seem quieter than other podcasts then this may be why.
The reason being that the loudest sound may be considerably louder than the rest of the conversation, and if you just amplify it so the loudest sound is set at -3db (Normalisation), then you may not get much amplification, (no amplification if the loudest sound is at -3db). Put simply, compression makes the louder things quieter, and the quieter things louder, evening it out so you get a more consistent volume level.
Apologies if I'm teaching grandma to suck eggs here.