He raises a good example for the age old excuse that “I don’t have time to write unit tests”. The example provided was some code that processed a data feed/push with transformation. In this type of scenario, it can take a substantial amount of time to complete a full test run. If the job takes an hour or so to run, that time can add up as bugs are found and the code is troubleshooted and retested. It would be far preferable to invest much of this time up front in creating and running automated unit tests which can simulate troublesome data issues and problems in a matter of seconds or minutes, in a more controlled and granular manner (than processing a full data feed, or even smaller simulated data feeds).
The fact that these unit tests can be fully automated in a Continuous Integration build as either unit tests or more robust integration tests, is even better. This way, the unit test is generated once, and will be run every time a build is done (probably at least once a day, if not multiple times per day).
If developers are not spending time writing automated unit tests, it very well may cost more time through the course of the project.