Specialized rear shifter issue - sorted

I have read a lot about S-Works bikes having rear gear shift issues so I thought I would share my experience which was ultimately resolved.

The bike was a Tarmac 2016 with 2x10 speed Ultegra. It was in very good order with correctly aligned internall routed gear shift cables (California cross) no visible cosmetic damage and about 1000km use.
The rear shifter would mostly change cleanly from cog to cog in any direction but would miss a shift when returning from either the large or small cog.
I did a visual check of the rear mech hanger plate which looked straight (As an ex-draughtsman I have a very good eye for angle but I would be the first to admit that this is no substitute for a dropout alignment gauge). The Wippermann chain was less than 75% worn and the cassette had no significant thinning of the sprocket teeth. It also was the correct length

Resolution:

1. I suspected there was a cable compression issue so I replaced them with Shimano SP41 outers and stainless inners. I have quality alternatives but experience with other bikes have taught me that some Shimano systems MUST be fitted with Shimano cables. I also removed the inline cable adjuster for the rear cable as one is present on the mech (I suspect this may not be a standard fit).
   I ensure the outer cables were cut to a suitable length and that every cable end was ground perfectly true to provide the eliminate the possibility of bed in or uneven/offset pressure within the ferrules
   The result was good but not completely resolved.
   
2. I replaced the chain with an Shimano Ultegra item. Big difference! transmission much quieter, gear shift more precise.
   
3. I found that the dropout hanger was insecurely mounted on the frame. I don't believe this worked loose as it appeared that it had two potential rotational positions which, because of the nature of the design may well cause a longitudinal offset in either or both. This is purely down to the construction process and is poor design in my opinion. The hanger is held in place by two tiny grub screws which locate from the inner face of the dropout which is part the carbon fibre frame joint and into the (threaded) alloy hanger. The face of the joint is uneven and the grub screw countersink is poorly applied. On this frame the rear screw did not pass through the carbon at the correct angle. This, combined with a countersink on the dropout that was slightly longer than the hanger it was designed for allowed that hanger to rotate a few degrees. The hanger was slackened of a quarter turn, enough to rotate it back where it was possible to re-tighten the screws by a little more than half a turn, indicating that the forward rotational position resulted in an interface between two surfaces that were not exactly parallel. Sure enough this last, obscure adjustment resulted in crisp, clean gear changes across the block on both chainrings.
   Job done.