I've used Athena really effectively to solve similar problems. If your data storage is relatively small and/or your queries relatively infrequent, JSON can be a good fit. As one of those dimensions expands, you can decrease costs/increase performance by converting to Parquet and compressing.
