Not every problem needs regex. For simple tasks, string methods are faster and easier to read.
When checking whether a fixed string is present, includes or indexOf is faster than regex and makes the intent clearer. For checking a specific prefix or suffix, startsWith·endsWith are just right. Using regex for such simple checks only makes things harder to read.
Splitting or replacing by a fixed delimiter is well handled by split or replaceAll (with a string argument). Regex only shows its true worth when there are multiple delimiters, when repetition or conditions are mixed in, or when the pattern is variable.
Regex is a good fit for format validation, token extraction, complex replacement, and matching several candidates at once. Conversely, simple containment checks, fixed replacement, and fixed splitting are better done with string methods. Fit the tool to the problem.
The criterion is simple. Ask yourself, "Does writing this as regex make it harder to read?" A regex your teammates can't decipher becomes a breeding ground for bugs. In many cases readability matters more than a little performance.