Answering the "why" questions is critical. If there is not a defensible answer, substantiated by some mesurable business need, then either seek the answer or don't change a thing.
Examples of good reasons:
- Get feedback from the product owner before releasing to the customers
- Reduce the risk associated with monolithic deployment
In some cases releasing something early doesn't make sense. For example, the software that drops the ball on Times Square at midnight at the start of a year doesn't need to be released in increments to the customer. It may be developed using agile processes, but it doesn't need to be released to the customer in chunks.