Sadly, many manufacturers have been pushing "USB mics", which bundle the preamp and an analog->digital converter into the mic body, and they've been very successful at marketing/selling these devices.
I say sadly because these devices violate rule #1 of digital audio: there should only be 1 sample clock. The moment you start using these mics in combination with any other digital audio stream (e.g. playback via the builtin or some other audio interface), there are now at least 2 sample clocks, creating the requirement that some software layer does resampling to keep things in sync.
Much better to just get a cheap "proper" audio interface, skip the USB mic option, and use a "real" (analog) mic.
I leave the condenser/dynamic question for someone else.
https://www.ebay.com/sch/i.html?_from=R40&_trksid=p2380057.m...
More or less anything from Focusrite, Presonus will serve you well. They'll get the job done until you care about subtle details (which may never be necessary), and might still be fine even after that.
If you're using Linux, the only thing to be sure of is that the device is described as either (1) "class compliant" or (2) "works on iPads" or (3) both. Linux users have much to be grateful towards the iPad for, in particular the ban on drivers meant that USB audio interface makers really had to get their act together to ensure that their devices worked with a generic USB audio driver (just like the one on Linux).
But yeah, for most folks a dynamic is fine, cheaper, and more sturdy. Or a Blue Yeti USB mic. My point wasn't to drill into the details and offer a recommendation, though, other than "wire your stuff, and you'll be loads ahead". What is on the other end of those wires is a separate discussion. And I'm probably not the one to lead that discussion. (I'm a musician, not an orator.)
Half my team use speakerphones and it drives me nuts.
The problem is that, just like cameras, size matters for mics. In order to even sense the full frequency range of the human voice it helps to have a larger diaphragm size. For headsets they range from microscopic to tiny which results in the voice recorded with the headset sounding like "telephone" voice (ie it's designed to capture the most important frequency range of the human voice, the medium frequencies) however, to me that sounds bad, artificial, cheap so I wanted more. Capturing more of the frequency range allows you to sound more natural, you know, like you are actually there in the same room talking to that person.
Depending on your usecase this may or may not be a consideration.
EDIT: as for noise issues, it's always best to avoid recording it if possible but even then there are software techniques to remove it post-facto. In order to achieve the former you can get super/hyper-cardioid pattern microphones that are very sensible recording sound that comes directly in front but not from other directions. However, no pattern will save you if you use a condenser microphone in a noisy room. It may be possible to use a second mic positioned away from your mouth and use software to remove the noise by essentially computing the difference between the two (some headsets can do that). Alternatively you can go for a desktop dynamic mic, just make sure you speak loud enough, but at least you won't get noise :)
It is, because users don't have to worry about mic technique or having a studio-quality room.
Anyway, if you've got a headset you're happy with and no one complains, I'd just use that if it were me. Again, I've got the "good stuff" lying around already, so I just use that. I would not recommend anyone else go that road unless they have other uses for the gear.
Do you use some USB interface that accept a separate microphone and a separate headphones jacks? If you do, any advice on what to look for and what to avoid?
https://motu.com/en-us/products/m-series/m2/
It's overkill for online meetings, in that many of its features will go unused. It's a well-built box, though, with better build quality than the (cheaper) Presonus it replaced. If there's just one mic or input, and some headphone output, then something like this would be less expensive (and less to fiddle with):
https://www.sweetwater.com/store/detail/UM2usb--behringer-u-...
That's the first single input, inexpensive audio interface I found on Sweetwater; just an example, not a recommendation.
Add to that a cheeeeaaap dynamic mic:
https://www.sweetwater.com/store/detail/XM8500--behringer-xm...
Now, for under $100 (surely you have some wired headphone lying around) you'll sound better than 95% of everyone else in online meetings. That's why I have no recommendation or ones to avoid, because for online meetings all it has to do is what it says on the tin to be better than what you've got now, and "good enough" for online meetings. If you're into music, then we can talk about which one is better over another (and frankly, I'm not the one to ask).
Now, that's a lot of money and fiddling. Bang-for-the-buck IMO is the already-mentioned Blue Yeti, along with a wired set of headphones (anything, just wired, and not coming out of the speakers to take the load off the DSP feedback cancelling). So, prolly $150 total? Point being, if you use just a wired mic and headphones of any quality, you're loads ahead in sound quality of those using the built-in mic and speakers on their laptop (or worse, their phone).