What you trying to achieve is considered as a very, very bad approach to tests automation. You should avoid making dependent tests. Basically you should have a total control over data input in your tests and over what happens to the test data during the test. What you shouldn't have control over is, obviously, test result, which you should check in the end.
To put it simple, consider getting test data ready as a pre-conditions step which should be fulfilled before the test run. Also, you should be certain about what you do to that data in tests. Otherwise, how can you be strictly certain about what do your tests test?
There are 2 common ways to get test data ready: either using AUT's API or injecting it directly into database. The second one is more idealistic and has multiple implementation options, though, it might be not available on your specific project.