Bringing agility to big data applications can be challenging for several reasons:
1) There is a very long feedback cycle when developing workable software applications due to the time needed to wait for quality data.
2) Switching costs between tasks and projects are high for big data applications.
3) Releasing features to all users can result in more questions about data quality issues.
4) Successfully developing workable big data applications has a very low success rate and wastes many resources.