I'm quickly complicating things and this is almost certainly flawed, but: payout multiplied by acceptance rate minus submission fee equals return on submission. But again, for publishers with a small sample size of submissions in your data, that's bound to be misleading