I made an aswer to a similar question that was asked a while ago. However, since it is posted in a restricted category, I don't know if you have access to it or not. Since I think I would give essentially the same kind of answer, therefore, I'll simply copy it here.
Here's the links
Part 1: https://forum.manjaro.org/t/old-manjaro-stability-question/84561/13?u=frog
Part 2: https://forum.manjaro.org/t/old-manjaro-stability-question/84561/15?u=frog
I know it is a really long post, but I consider that it is important that you have the right expectations for Manjaro and have all the information you need in order to decide if Manjaro is really the right distribution or not for you. Maybe it will also help you to understand better what is Manjaro, what is its purpose and its goal by reading the perspective of a user of a rolling-release, cutting-edge Linux distribution.
A really long answer for a question that seems, at first glance, to be not so complicated.
The answer of that question is not a straight up yes or no; it is a lot more nuanced and there is many factors to take into account.
Manjaro is a rolling-release distro that aims to provide the lastest software available from upstream in a reasonably fast pace. What it means it that you will get rapidly the lastest features from upstream, the lastest bug fixes from upstream... and unfortunately also the lastest and newest bugs from upstream too. This is a risk you essentially have to accept on Manjaro. Because of this choice, inherently, Manjaro cannot pretend to ever be more stable than classic fixed-release distro like Debian or Ubuntu simply because the environment change a lot and often. Stability is of course way easier to achieve when you do not push new versions all the time, but only when there is a new version of the distro, which is what classic fixed-release distro are doing. If stability (stability as not changing much over time and stability as not being prone to regressions over time) is the absolute #1 priority, Manjaro may not be the best choice; you would most likely prefer to use a fixed-release distro and use the same version of each program for a long period of time instead. I think when Manjaro became more "mainstream", many people came here from Ubuntu or Mint and expected the same kind of logic than on their old distro; but in reality, Manjaro is a very different distro that will appeal a lot to some people and not necessarily be the best choice for some people.
Users is definitely a huge factor for the stability of a Manjaro system. There is many, many people who, I would say, overcustomize their system, and some people will customize their system not to fullfill a need, but for the sake of customization and because "Linux is customizable". The problem is that the more you will customize your system, the more corner case scenarios you will introduce on your system that will most likely never be tested and therefore, the more your system is prone to regression. On fixed-release distro, customization may be able to function correctly for a long period of time because software doesn't really change much, but on Manjaro, since you may receive a new version of a dependency with breaking changes in it, customization are more prone to break, and more often.
In addition to that, using software from other sources than official repositories (whether it's AUR, third-party repos such as herecura, GNOME website for GNOME extensions, etc.) will also add corner case scenarios that are unlikely to be tested. Anyway, Manjaro team won't really take into account software that comes from outside of Manjaro repos when brewing incoming updates. If you only use third-party sources for applications, it might not be that bad (at worst, your application stops to work, but your system in general should be okay). If you start to use kernels, init, system libraries, or even extensions for DE, I would say that you should be prepared for the worst (from unusable graphical session to unusable system completely). I say that because I saw many people blaming Manjaro to break something that doesn't come from the official repos and therefore, has absolutely no control on.
Many users also don't maintain their system very well. We have seen for example people who keeps downgraded packages for so long that it ends up breaking something, ignore updates for some packages and then forget about it until it ends up causing an issue, people who only do updates once every 1-2 months and then have trouble to figure out what went wrong with such big system upgrades, people who tries to install packages on a out of date system and don't understand why some packages can't be installed or ends up breaking their system with a partial upgrade situation, people who keeps EOL kernels that won't receive any updates from Manjaro anymore and not being able to do their system upgrade because of dependency problems (for the last one, those people are mostly people that want to use non-LTS kernel, but don't do the required maintenance for kernels that have very short lifespan), etc.
It has happen in the past that Manjaro Team asked to users to do special manipulations for particular update set, for example doing a system upgrade within a TTY or forcing the package manager to downgrade packages. When Manjaro Team gives special guidance, it is serious and must be followed, or else at best, you won't be able to start the system upgrade and at worst, you may simply break your system completely. However, we all know that not everyone will read the news and will update as they usually do, only knowing that they had to do something in particular when it is too late, if they even bother to read the news before asking for help. For those kind of people, they will blame Manjaro to be unstable and breaking all the time; for a significant part of the community, they will blame the user entirely for not reading the news before upgrading because if they did, they would not have broken their system at all. In my opinion, I think that Manjaro Team should avoid manual intervention as much as possible and unfortunately, they sometime don't do anything to avoid manual intervention even when it is 100% possible to do so. However, there is moment when manual inteventions are unavoidable unfortunately.
It can happen that Manjaro Team f- up, as it can happen to anybody. A subset of "Manjaro Team f- up" situation IMO is when Manjaro Team wants to be "good guys" and revert back to an older version of a program because many people complains on the newest version and it just ends up backfiring on us later. Examples of this subset of "f- up" includes Winter 2019 when many people ran systemd 239.300 while 239.6 was in the repos and a rebuild against libidn2 was needed and therefore, users had to do a manual intervention to force downgrades from 239.300 to 239.6 to keep a bootable system (which is an issue that happened because any update of systemd to something higher than 239.6 could make the graphical session crash in the middle of upgrade, causing breakage, Manjaro Team went back from 239.300 to 239.6 and tried to wait for a more stable transition, which ends up never really happening, so they did the upgrade from 239.6 to >239.6 twice instead of once); there is also the "Mess-a" in April, when the team put mesa 19.0.1 packages in the repos in a update set, but put mesa 19.0.1+really+18.3.5 in the repos previously because many people complained that Mesa 19.0.1 was too buggy on their system, going on mesa 19.0.1 was supposed to be optional, but mesa 19.0.1+really+18.3.5 became totally broken with the update set that put mesa 19.0.1 package in the repos, most likely I think because LLVM 8.0.0 (which Mesa depends on) was introduced at the same time. So by doing the "good guys", it happens that we end up having more problems in a period of time than if we went for the "deal with it" approach in the beginning, especially if it is not handled carefully. The "deal with it" approach is what Arch Linux did (they never went back on systemd, they never went back on Mesa too). The problem is that if you go with the "deal with it" approach, people will then say that Manjaro breaks all the time and the Team does nothing to fix their sh-.
It can be a factor. Generally, doing updates in a graphical session is okay and should not be a problem most of the time. However, when the devs recommends to do an update within TTY for a particular update set, it is serious and you should really follow the recommendation. Unfortunately, as I said above many people will not follow the recommendation (either because they didn't read the news, or didn't care) and ends up with a broken system.
Nothing really wrong with your way of updating your system IMO. Personally, I exclusively use Pamac GUI in Testing branch (because it is convenient for me, I like this package manager a lot for the most part and hey, someone has to test it out if it is solid, right?). If there is something special to do for a particular update set, it will be mentioned in the announcement.
A mix of both I would say, for all the reasons mentioned above. More specially, Manjaro not being stable is not entirely because Manjaro Team does a poor job at maintaining their distribution, it happens really often that software itself is just buggy in itself and only upstream can really fix the problems.
Since software provided in mesa 19.0.1+really+18.3.5 packages was broken, it means that people who kept this package had a lot of trouble with programs that rely on OpenGL or Vulkan if they use the Mesa implementation (which is very popular with Intel and AMD GPUs, most people with a NVIDIA GPU will use the non-free drivers made by NVIDIA instead). The program that needed OpenGL or Vulkan couldn't work properly entirely or partially when you were using a broken Mesa. The exact side effect varies.
A lot of confusion since the first statement made by the Team said that both mesa 19.0.1 and 19.0.1+really+18.3.5 were working equally well so the package downgrade from 19.0.1+really+18.3.5 to 19.0.1 was supposed to be optional, but it ended up not being true in practice considering how troublesome keeping 19.0.1+really+18.3.5 was.
But now that we have Mesa 19.0.2, all of this is solved.
The kernel is not the only part that can make a system non-functional though. Being on a non-LTS kernel is not a problem if you check up for EOL kernels and uninstall them when it reaches End Of Life, which is the thing many users running non-LTS kernels don't do.
Special guidance for update set is something rare though. However, I prefer to be honest and warn you that it can happen so you have the right expectations from Manjaro. If I told you that Manjaro absolutely never breaks and you can always do whatever you want blindly and nothing will happen, I would be lying, and I would most likely lying for like any distro.
Also, I do not know what distros you have used in the past (and if you have used any sort of Linux distros in the past), but from what I have seen on this forum, many people that comes here from, let's say, Ubuntu or Mint, tries to use Manjaro a bit like if they were on Ubuntu or Mint and expect the same kind of behavior than on their previous distro, compares Manjaro with their previous distro with the point of view and logic from their previous distro, tries to do the same things the same way as in their previous distros and ends up being disappointed. The problem in that is that Manjaro is an entirely different distro, based on an entirely different distro too, with it's own goal, it's own way to do things, it's own logic and reasoning behind technical choices, etc. That doesn't necessarily make Manjaro bad, it may simply means that Manjaro is not a good choice for that person to begin with.
It can be a good idea. Anyway, there is generally a big update set every 1-2 weeks on Stable branch, so you have plenty of time to plan your system upgrade and keep up with repos. Do not feel rushed because you are notified that there is updates, but do not neglect doing updates regularly too.
Also, if you frequently go on the forum, you will definitely be more prepared and have an idea of what to expect.