Reality check: you are talking about a difference of 0.02mm over 5mm for folks that still use inches that is 0.000197". I would suspect that you can't really measure down to the hundredths of a mm range accurately and your chasing faulty measurements. To get an accurate believable measurement you must be able to measure at 10x that resolution. So to be quoting measurements of 0.01mm you need to be able to accurately measure 0.001mm. When you get to that scale you have issues with the measurement tool construction, springiness of the tool needle, dust under the needle, etc. etc.
If you know your steps/unit is 640 based on your lead screw pitch, drive reduction etc. then type in 640. You are NOT going to make it more accurate overall by trying to measure distance traveled to one point and altering the steps/unit. As you move an axis along inaccuracies from the lead screw pitch, drive reduction pitch errors, the table shifting slightly in the gibs all come into play. Let's stay you have a table with 200mm of travel so you tell your machine to move 100mm and measure the distance travels with an uber accurate measurement device. Now if you only move from your starting location to that same position that you calibrated to you will be very accurate as long as the temperature never changes, and nothing ever wears, etc. Your calibration did not make your machine more accurate overall.
What I'm trying to say is don't chase your tail trying to achieve accuracy you can't even measure. When you are actually machining you will have far more inaccuracy dies to tool flex, your machining strategy, tool wear, etc.