1. Immerse yourself into the data trough reading it thoroughly
2. Generate a first set of codes
3. Apply, refine, elaborate, split or group codes
4. Test codes for reliability (trough comparison with other coders)