- Hey X, I brought you into today to talk to you about something I noticed. There is an issue/errors with your work, when doing Y[provide evidence of common errors/average errors of others], I just want to say you're not in trouble, but we want you to improve this area and wanted to make sure you were aware of it.
- I want to extend any resources I have available for you to improve in this area, and of course I have some ideas, I wrote everything down on this paper/email.
- You can come with a plan of your own. Or we can collaborate on it--if you're not sure why issues are happening. Take some time and think about it, and when you're ready to talk about how we might ago about improving in this area, please schedule some time with me and we can work on this together.
Stick to facts. Clearly state your expectation, and show how they are not meeting that expectation. Then place the ball firmly in their court. The goal is not fixing the issue for them, but getting them to take responsibility for fixing it themselves.
Refrain from creating a "shit sandwich" by putting the critique in the middle of praise. That makes the conversation ineffective. These conversations are never fun, but they are important to have, and you eventually get used to it.
It's not about berating anyone. It's about constant subtle modifications. They compare it to driving a car. Even if you're on a completely straight road, you can't just keep your hands off the wheel and expect the car to continue going straight. You provide regular nudges to keep the car on the road.
Maybe this analogy doesn't hold up that well anymore with auto-correcting cars :)
I think your concern about not providing stronger feedback is valid. Highly recommend the series of podcasts they have around this topic. You'll think about feedback in a much different fashion.