Dell XPS series looks fantastic from the outside, but there are a few things that stop me from wanting one, that I've mostly observed from other owners of XPS laptops, having not owned one myself.
1. The keyboard. It's apparently not so bad once you get used to it, but compared to typing on my Thinkpad I find myself unhappy with it; it feels spongier in comparison, and the key travel is pretty short (though not nearly as bad as the Apple butterfly switches.)
2. The docking situation. Thunderbolt 3 docks are not really that appealing to me because it becomes difficult to drive many displays at 60 FPS with them. In all fairness, the Thinkpad workstation dock can also only drive 2 displays 60hz 1080p OR 30hz 4k, but at least it's reliable, and my Thunderbolt port is still free if I want more displays at the cost of convenience.
3. Linux support is actually really awful. The XPS 13 developer edition is supposedly an exception, but I have heard that it's really only well supported with the exact version of Ubuntu it ships with, and that updates have occasionally broke the system. Admittedly, the same can be said for some Lenovo laptops. My P51 is pretty bad under Linux (though it's gotten to the point of being usable, thankfully.)
4. Reliability. A coworker has replaced his XPS like 2 or 3 times now, and one of the chargers also died. I hope that's the exception. Obviously even Thinkpads are not as rock solid as they used to be, but even though I found this level of failure to be pretty extreme, it lines up with what I've witnessed with other Dell laptops and is one of the primary reasons I've never owned one.
Honestly, the choice of laptop comes down a lot to priority, but I feel like XPS exceeds mostly in the categories of portability and style, two things that I do not value nearly as much as robustness and performance.
More than anything, the most important thing you can do is research the crap out of any option you're considering buying. I've found that Arch Linux wiki pages about laptops are quite useful, with the caveat that you have to be careful to not assume two similar models will perform similarly since that is often not the case.