Have you shot the Shield in your
other hand? Did it sill shoot that far off in the
same direction? If the direction of the POI "changes" when you change hands, it's likely not the gun or its sights.
The Shield is a rather thin single stack, with a short reach to the trigger. It's possible that you're getting too much, or too little, finger against the trigger, and then are unknowingly pushing/pulling (as the case may be) the trigger - and therefore the muzzle - off to one side at the
moment the trigger press is completed. That moment is when a lot of shooters may "lose contact" with their sights, so they don't see that their sight picture has actually changed at the moment of firing. Then, they blame their sights.
Oh yeah, that weird kink in the mag spring has happened with some other mag springs that transition from small to large coils. Annoying, to be sure, and apparently something to do with the manufacturing (heat treat?) of the springs.
You could try some of the old remedies for diagnosing shooter-induced POI shift ... rest a nickel on the front sight (and see if it moves a bit to the side on the sight post, not just falls off), or get some actual Dummy rounds and have 1 or 3 loaded into some different mags (by someone else, who also loads the gun, so you don't see the top couple of rounds in the mag during loading), and see what happens when the dummy round comes up.
Dummy round drills only work if you actually think the gun is going to fire a live round, though, as a lot of people unconsciously change their trigger press to "make it good" when they think they're about to be fooled by a Dummy round being chambered.
I usually take a shooter's pistol and unload it, manipulate, it, examine it (turning away from them, so they can't see exactly what I'm doing), making them think I'm checking something on the gun ... while simply making sure the mag isn't seated to let the gun chamber a live round at one point ... and then hand them back a gun which I tell them is loaded and ready to fire, and instruct them to make a precision shot on the target. That virtually always reveals a flinch and/or trigger clench or jerk that pulls the muzzle offline as they fight the non-existent recoil or yank the trigger off to one side at the end of their trigger press. In other words, they really think the gun is going to fire, and they do whatever it is they're doing that interferes with them getting a good hit, which is easier to see if the gun isn't recoiling and cycling.
In order to
correct a shooter problem, first you have to
prove to the shooter that it exists.
If you send the gun back to the company, they'll check it for a barrel/slide problem and test-fire it, but if it's not a gun "problem" that won't diagnose the
nature of the shooter problem (other than it being a shooter problem).
In order to correct a shooter-induced problem, it has to be diagnosed (while getting the shooter to acknowledge that it's related to something they are, or aren't, doing).
Can't be diagnosed and corrected online, though, but has to be observed and resolved in-person ... which is why they don't have me telecommute to work range sessions.
Best of luck in figuring out what's going on. The Shield is still commanding the market in the little subcompact single stack segment. I just learned that the company has put the 1.5
million unit mark in their rear view mirror, and production & sales are still hot. It's apparently broken all existing company sales records, for any handgun model in their history.