There are a lot of variables involved in video conferencing.
Lag is most often caused by bufferbloat outside of both the service and the users' systems, and consequently the alternative is not less lag but intermittent dropouts or lost connections.
Cheap solution: Tell users to use a "real" computer (not phone that may randomly switch network quality) and use a reliable connection (not cafe wifi or a cellular network).
Expensive solutions: Buy dedicated conference gadget for all users and hold conference calls using some service that require that gadget, forcing users to use their real computer, which is likely on a reliable connection. Pay all users for a reliable connection. Pay all users for a VPN that only works over a specific reliable connection, and use a service requiring VPN access, forcing users to use their reliable connection.
Room noise is caused by microphone being placed too far from the mouth.
Some software tools might have luck compensating for that.
Cheap solution: Tell users to always use headset (not microphone built into laptop). This must be said out loud and when the issue occurs and repeatedly, not just in a memo before or after, because the problem most notably affects others than the user causing it, so there is a lack of intuitive correlation.
Expensive solution: Buy dedicated conferencing microphone+loudspeaker for all users.
Echo of your voice is caused by another user having bad isolation between their microphone and loudspeaker.
Like for room noise, some software tools might have luck compensating for that, and some dedicated conferencing hardware containing both microphone and loudspeaker may more efficiently compensate by careful placement of those devices coupled with software calibrated for that exact hardware (e.g. knowing the exact distance between them and amplification of them).
Solutions are same as for room noise.
Gap in someone's speech is more often caused by too little lag, and less often by a loose connection in the wiring of the microphone.
Solutions are same as for lag - and also buy high quality headsets for all users.
In addition to each of above causes in isolation, they are all amplified by users connecting with varying software (versions, web browser providers, operating systems, computer designs).
Cheap solution: Tell users to update their systems and web browsers, and use "real" computers (not phones).
Expensive solution: Use a service that discriminates users of some systems and hardware types, effectively forcing everyone to use fewer combinations.
Summary of above: Choosing one service over another may help reduce amplification of frustrating issues, but rarely help address the underlying issues themselves.
If most of your users have iphone, then tell the rest to also use and iphone so that you use similar hardware and software and tools.
Same for Windows vs. MacOS vs. Linux.
Same for Firefox vs. Chrome/Chromium vs. Safari - and only browsers on "real" computers count here: Products on phones are different reduced ones!
In the end, the only factor that is really truly tied to the choice of video conferencing tool is your need for dial-in from a landline phone.
Technically that means a solution that supports connecting with SIP, and a service provider running a bridge between SIP and the global POTS network.
I see no benefit in the various closed-source offerings, and recommend you to choose among solutions based on Free Software (a.k.a. Open Source). But since you require dial-in from a landline phone you should not expect to find any free-of-charge options (because landline connections are often costly).
These Free Software video conferencing platforms support SIP dial-in:
Other Free Software platforms and tools work with SIP as well, and service providers might use varying combinations.
Personally I would not use a "platform" but Janus
that is far more lightweight and efficient than any of the "platforms".
Which services using those platforms and offering dial-in, you need to locate yourself: that requires knowledge on which country code you need for the phone number.
