I'd suggest to make a bit higher detalization than just 3 zones (as stated above). 

You probably have various yield limiting factors in different areas. 

Target yield approach is good, some of our clients is it with equation-based zones.

Each field is unique, some recs based on our clients' cases: 

1) Analyze field:
- Field potential maps, check various years, is the field stable from year to year? 
- Clean/calibrate Yield maps if you have any, to use in 
2) Define management zones logic: the same zones for seeding, fertilizing and CP or different per operation? 
Some of best Practices (based on GeoPard clients' e):
a) VR seeding based on Multi-layer Field Potential (good idea here is to aggregate historical vegetation, topography, soil sampling & yield data; Remote-sensing derived proxies for Organic Matter and Moisture)
b) VR Fertilizer based on Equations: soil data and Uptake calculations
c) Crop Protection using latest vegetation
3) Define the logic of creation of zones. Factors to consider: field potential, soil properties, topography, yield data
4) Design rates strategy: granularity with agronomic formulas like in the example below or take avg +-10-20% in higher/lower zones. 
5) Plan Field Trials  
6) Do post-harvest analytics. Calculate Profit Maps, Fertilizer Use Efficiency. 
7) Improve agronomy for the next season based on Post-Harvest Analysis. 
8) If you don't have historical yield info - create it with Synthetic yield maps module