Ideally, I would prefer to hand pick the resources on my team so I can at least fool myself into thinking I am off to building a high performing team. In reality, you get what you get. One thing I have often observed in my 15+ years is that, once a resource is labeled "bad performer," they have a hard time shaking it. I might argue that they performed poorly not because of their lack of capability but rather the environment in which they were placed. Place them in another environment and you could get much different results. I have noticed that with my own performance.

Another issue I have observed is misalignment. Resources are often placed in roles outside of their expertise or level of competency.

As the PM, I would carefully consider what role best matches their skill set and capabilities and align them with that. Establish expectations up front so the resource knows what to do, when to do it, and how. If necessary and I could afford it, I might also try to have another resource shadow the poor performer to help coach. Period check-ins would not hurt, either.

Of course, at some stage you have to cut bait. If nothing else worked, then I would push for a resource action in order to protect the project's performance.

One concept that I often think about is the Apollo Effect. Sometimes, a team of the best and brightest does not make the best team. That is probably true more often than not, in fact.

If you worked with a guy and marked him as a poor performer that's important information for you. It's better that way than if you had to learn the fact in the middle of the project.

What can you do once a project is started?

1. Personally I wouldn't do any special "new opening" stating that we worked together and I consider the guy as below-average performer etc. It might be just a bad time for him or hi might have done huge improvement from that time or whatever. The only thing I'd do at the beginning is keeping the guy out of the critical parts of the project.

2. I would monitor his work from the beginning but would do it rather softly. Nothing like regular code reviews or verification by the other team member you trust. At least not as far as this isn't a rule applied to the whole team. I'd rather check his part of application by myself (if it was a working piece of application) or put his code at the beginning of testing or give him other easy-to-measure tasks like setting up build server creating some supporting tools for others or something. This is basic risk management for a risk called "that guy may be a poor performer."

3. Given that the person is still a poor performer I should see some early alarms. Then I'd do what probably most of us would: face-to-face discussion about the problems I see and ways to deal with the situation. At this point it is fair to add some formal verification of tasks assigned to that person.

4. If the situation didn't change I'd repeat discussion but this time I'd make it clear this is the last chance.

5. If the situation didn't change I'd start replacement procedure, which is heavy-dependent on the organization we discuss. It might be asking my superior for a replacement or talking with a functional manager of the person about his performance or having some informal discussions with a person superior for me and the functional manager or preparing some metrics which blatantly shows who hampers the project. Anyway once I'm convinced the guy brings more harm than value to the project I'd work consequently to get him removed from the team (and fired if necessary) using any tools I had. This isn't a pleasant part but sometimes has to be done.

This is how I'd act in ideal situation. In reality we often don't have enough time to wait for each "test" to show its result but if I have doubts I'd interpret them to the poor performer advantage.

You need to mitigate the risk. You can see if you can get them replaced. Minimise any and all deliverables so none impact your project. Review their work often. And yes, when appropriate, speak with their managers or HR...

If the person is 'officially' marked as a poor performer than I would work with HR to ensure we are tracking all potential issues - aka 'managing the person out'. A bad match between person and role is not good for either the person or the team or the company. I would create a risk-mitigation plan to ensure that any miss-steps by the person are minimized and provide what mentoring/training is available to help the person improve (if possible). In most companies you are not given the choice of resources, but you are given the choice of recognizing any/all issues and communicating out the plan to address and the potential impacts.

It's a risk. You should raise it privately with the client/sponsor and have a plan to manage the risk.

You may accept it, you may mitigate it, etc.

Keep in mind people and politics often come into play, but keep your boss informed.