The whole Kafka Connect platform is built around json and jsonschema, being the de-facto standard.
Confluent (creators of Kafka) created the schema registry, which allows to have Avro data in Kafka, where the schema is stored externally in the schema registry, cf http://docs.confluent.io/3.2.0/schema-registry/docs/. The payload is therefore way smaller.
The added benefit is that we can have strong enforcement of schema migrations and compatibility guarantees, such as backwards, forward, etc. Once this is implemented, people can start using Kafka Connectors (https://www.confluent.io/product/connectors/) so they can externalize the Snowplow data to ElasticSearch, S3, etc, for basically free.