The calibration for PH meters usually employ something of a straight line curve with the 7.0 being at the center (0 mV). Any voltage at PH 7.0 is an offset.
The slope and offset of the curve (a straight line) will determine the readings at locations other than 7.0. Depending on what the values are stored at the other calibration points, let say 4.0 or 10.0, the slope of the line can differ.
For instance, here is a graphic of where they take two far end points to correct calibrate the meter/probe combo. Correction is the dashed line.
You can imagine a case where one of the points acts a bit differently causing the slope of the corrected line to change. The value at the 7.0 might be accurate but as you move along the slope to different PH values farther away from the 7.0, the error increases.
There could be some different reasons perhaps, thinking out-loud, 1) old or contaminated calibration solution 2) something in the solution is interfering with the PH chemistry of the probe 3) one of the points on the calibration curve is incorrect. e.g. need to re-run full cal. 4) temperature sensor is not calibrated or temperature has not stabilized when reading 5) probe needs to be cleaned or replaced.
Note that for three point calibration, if the manufacturer is using a straight line curve for compensation, two points define a line. Now, with three point calibration, this can actually throw off one end of the PH scale if all of the point do not naturally fall on a straight line. But, they should be close.
One might try changing the order of the calibration if doing a three point cal. This entirely depends on how the meter calculates the compensation which I don’t know. If they recalculate on each new data point, it “may” make a difference at least for one-half of the curve. For instance, if the order was 7-10-4, try 7-4-10.
Here are some general thoughts on what can effect the accuracy of PH probes / meters: