Sometimes it is enough to get optimum exposure and sometimes, just like any reflective metering, it doesn't produce the exposure that you want, but the evaluation is linked to AF point, and as Sharlin says, it's easy enough to watch it happening. You're not on your own though in not realising that's it's been happening. I recon the majority of users don't realise !!Not saying it doesn't do anything. Just saying that for my photography it does not work. It may prioritize the focus point but it does not prioritize it enough to make a meaningful difference for me. Which is why i have tobuse either spot metering linked to center focus point or most of the time use centre weighted average. Evaluative rarely ever cuts it with birds against a bright sky
The metering and it's 'intelligence' is always getting better from generation to generation, and so I would expect the R5 to be better just as the future 5DV will be. It's interesting that Canon have chosen to highlight this in the R5 specs, maybe they too have realised no one knows about the feature. Certainly now, with my 5DS I really don't have to worry about getting a useable exposure, although I'm still one of those people that like to take an incident meter reading if feasible.
As I'm sure you are aware, Spot metering linked to a single AF point is a really dangerous way of metering and you have to be aware of its limitations, which is probably why Canon only offer it in the 1 series. Spot metering, by its very nature, is intended to be used over multiple areas of the scene to assess an exposure, not as a single spot reading, yet this is exactly what linking it to a single AF point is likely to do, and so a degree of EC is likely to be required anyway, otherwise you could end up with a right royal hash in overall exposure of the image.