There are several factors that might cause SAM's results to be different from measured data from a real system. For example, the weather data you use for SAM's simulation should represent the same conditions under which the real system operated. Also, you should make sure that the shading and loss assumptions in SAM are consistent with how the real system operates. And, you should be careful that the module, inverter, and system design parameters in SAM are the same as those for the real system.
You might find these photovoltaic system case study reports to be helpful:
sam.nrel.gov/case-studies
Best regards,
Paul.