I had some deep qualms about this -- "this cannot possibly have any scientific basis, right? What is the point of this? How can anyone possibly assert that there are four basic kinds of personalities, that they respond consistently, or that they way they interact matches consistently?"
A colleague's view was something like -- okay, yes, stipulated it's bullshit. But what's important is that we managed to get 100 people into a room together to talk about how they like to work with each other. The "colors" are canned, not-quite-true insights that can jump start the conversation about how people like to work. We gave people some shared vocabulary, however limited, for talking to each other, and they then spent hours working with each other to discuss how their communication and work styles have strengths and weaknesses. So it was a good team bonding exercise because it felt like people would try to work together more afterwards.
I'm still a little torn about using a likely-false concept as a pretext for people to get together and connect with one another -- but there's plenty of this in human history.