With respect, I suspect that what might solve all of your problems is some good old fashioned statistical quality control since it appears your companies have exactly the sort of digital data to do it easily. Statistical QC will distinguish between what is a natural deviation from the mean and an actual departure that indicates something is wrong with the piloting. In fact I'd be surprised if most airlines aren't doing this already if they have the digital data.
What gave me the clue is in the "stabilised by 390 ft RA" comment.
Are you safe if your data indicates that you were stabilised per whatever your company's manual say at 501 ft, but not safe at 499 ft?
Are you safe if your manual says 180 knots and you fly 178 or 182 knots?
Safety is not a Yes/No thing unless you can prove that what has happened is unsafe.
You cannot make any determination about whether a pilot is doing the right thing unless you can determine that he is doing something statistically different from his peers.
For that, you need to determine the mean and standard deviation of speeds and heights for each of your aircraft types at each "gate" and across your entire pilot workforce. I imagine your digital data can do that easily.
Once you have that information it is a simple matter to determine if non-conformance to the "book" figure is a matter of normal random variation or a statistically significant deviation from the rest of the Pilot population.
You can also look at the personal mean and standard (statistical) deviation for each pilot, and analyse it any number of ways to work out trends and spot problems before they cause anyone grief. Might also show a few surprises too.
Furthermore, if you really have all that data and can do the above analysis, then you ought to make sure your airlines share it between themselves and benchmark their performance.
It would also end the eternal arguments about who are "Cowboys" and who aren't.