There are a number of assumptions that go into the modeling results within the outdoor planner. I would need to see your KMZ output that you are comparing with your real-world performance to know for sure. More than likely though, you used a higher design margin and your environment isn't as bad as the margin chose, or your client devices that are using to measure your actual have better Rx gains (atennas, better radio sensitivity, etc) and/or have higher output radios than the standard assumptions in the default client devices.
In general though, the outdoor planner standards (unless modified during the generation) take a very coverage-conservative take on the modeling. It requires very specific information from the user to fine tune things like design margin and client device type. Hope that makes sense.