Another suggested starting point... "What are we (as a company) attempting to incentivize?"
A lot of what's already been stated by others in this thread will then 'naturally' come together once you start down that path of aligning incentives to the strategic goals of the organization. Goals (financial or non-financial) can be as diverse as one can imagine depending on where an organization is in its maturity. Once the goals have been determined, the next question is how do we (as an organization) get there? That's where metrics will come into play. And these will likely be different from business group to business group as each will have their own way to "get us there" as a whole. By then, the plan design will hopefully be a clearer picture -- one that could be individual or group or team incentive or a mix -- whatever best gets you to the finish line. The important thing is that everyone in the organization understands that they are all contributing to the *same* goal... helping each other... working together... and that you are not creating incentives for employees or employee work groups to work *against* each other. Obviously, that would be incentivizing the wrong behaviors and working against your organizational strategies. This is one reason why many incentive plans are designed as a mix of multiple 'factors' or 'multipliers' that are both individual, group, and team based. Each contribution is calculated separately (based on results in each area) and then the final result determines the award or incentive. To take this one step further, each 'factor' would have its own separate threshold for "success" -- and depending on what you want to achieve with the plan -- thresholds can significantly reduce the final award/incentive or eliminate the award/incentive altogether.
So... I guess what I'm trying to add to the discussion is the strategic planning piece. If you have a strategic planning department that's an ideal place to start. If not, and you're a smaller organization, then a visit with the company president probably isn't a bad starting point.
Just more food for thought...