When a new aeroplane design is certified (I'm including pretty much everything except for homebuilts and microlights here) you usually need to determined PPEC (Pitot PEC) and SPEC (Static PEC) so as to determine airspeed measurement error (which is a function of TPEC - Total PEC = PPEC+SPEC) and altitude measurement error (which is a function of SPEC).
Errors then fall into two parts - the overall system error (which is generally a function of the shape of the aeroplane) and the individual instrument errors.
Looking at the airspeed measurement error - this is down to TPEC (which lends itself well to being measured by a GPS method, of which the triangle method is one of 4 that I'm aware of). But, at the same time, instrument errors apply. The ground test set only measures instrument error.
If you look in the POH, you'll find a graph of airspeed measurement error - which will have had to comply with JAR-23 requirements that over the range 1.3Vs to Vne the maximum error may not exceed the greater of 5knots or 5% (so it could be 10 knots out at 200kn and be quite legal).
So, all being well, what you see in the cockpit is the sum of those errors (which are specific to type) and the errors in the instrument (which are individual to that gauge and were measured by that test set).
All being well, then IO540 what you are seeing in your GPS based calibration is the sum of that 5kn instrument error and presumably a 4-6 knot underread due to PEC, which you should see in the POH.
G