Sometimes it's hard to decide which way to go in the project.
2 techniques which help here:

1. Timeboxed experiments
2. Reversible decisions