we are only observing single listener (aka record observations) currently. Same approach that spring-kafka has taken. We can look at how to add that to the batch case sensibly once the dust settles on this basic use case.
Originally posted by @onobc in #146 (comment)