Dude, I used to just wing it with random junk, but then I spent *hours* debugging stuff that wouldn’t happen in prod.
Now I define sample data to match real-world distributions—like, if 10% of users are inactive, my test data should reflect that.
For big datasets, I’ll sometimes slice a sanitized prod dump. Just be careful with PII, obvs.
Now I define sample data to match real-world distributions—like, if 10% of users are inactive, my test data should reflect that.
For big datasets, I’ll sometimes slice a sanitized prod dump. Just be careful with PII, obvs.