I'm always surprised when I see people online complaining about Challenge Rating in Dungeons & Dragons. A deep roster of interesting adversaries and tools for balancing them for players is one of the biggest advantages D&D has over other tabletop roleplaying games beyond name recognition. Roleplaying games are so open and complex that I don't think it's possible to have a perfect system, but I believe that a tenet of good DMing is to know how to use CR— what situations to avoid to make it work.
Some would say that the CR system isn't necessary because balance isn't "realistic" and that analyzing enemy strength and choosing to run away should be part of the game for the players. If they enjoy that, sure, but to me it's like asking someone to play chess against three queens. Not all encounters should be perfectly equal to the party's strength, but it should be done intentionally, and CR helps— rather than hinders— creating those sorts of encounters.
Now, this is going to sound like I'm already walking things back, but the CR system has to be one-size-fits-all. It doesn't know whether or not you gave all the player characters +5 holy avengers or if they think that 30 silver pieces is a king's ransom. It doesn't know if your players are power gamers interested in maximizing DPS or if they decided to all be gnome bards for the lulz. It shouldn't take long for you to figure out that your party should be treated as a lower or higher level than they actually are and adjust accordingly. But again, you wouldn't have this frame of reference without using CR.
Where CR most often breaks down is when one side greatly outnumbers the other in a conflict. This most often comes up when dungeon masters want to have a "boss fight" against a single enemy who is much more powerful than the player characters. As tempting as this is, I don't think this is good design, even if it is balanced exactly as you want. Interesting combat is about giving your players interesting choices and one of the most fundamental ways to do that is to add in different adversaries of different types. Otherwise everyone takes their ideal positions and just starts hammering away.
By the same token, even using the up to 4x difficulty multiplier given in the Dungeon Master's Guide for encounters with 15+ monsters, it is still very possible to overwhelm characters with weak monsters that should be a balanced encounter. For a "normal," balanced encounter, I try and never have more than twice as many enemies as there are player characters. Everyone wants to recreate the battle of Helm's Deep from The Lord of the Rings, but the standard combat rules for D&D just aren't built for that and it tends to turn into a slog. Players are aware that the monsters are too weak to be a threat, so they conserve their limited abilities and just chip away at the horde for hours.
Fourth and fifth editions of D&D have largely limited creatures with "save or die," abilities with the potential to remove a character from battle based on a single die roll. Those that remain do so because the creatures are a classic part of the game's history, and also because this sort of mechanic— used sparingly— can be fun and exciting. However, it can wreck havoc with the CR system, introducing wild swings in difficulty.
I find that these sorts of creatures are balanced fairly well when encountered as a single part of a larger force, rather than with others of its own kind. By the math, three basilisks should be about as difficult as a group of five bugbears with one pet basilisk, but the first encounter has a much higher chance of multiple party members being turned to stone. I feel that the second encounter works much better. The bugbears increase the choices the players are forced to make, and you've still introduced the danger of being turned to stone, while somewhat limiting the worst case scenario.
You'll have to check through the Monster Manual and learn which creatures have these sorts of abilities and which don't. Fourth edition assigned roles to each monster that did a bit of this work for the dungeon master, but it also added to the system's reputation of being too much of a tactical combat game rather than a roleplaying system.
So yes, I'll admit that the Challenge Rating system isn't perfect, but it has been play-tested and tweaked to the point that it's as reliable as it's going to get. Intentional or not, I even feel that some of the pitfalls that lead to making an encounter that isn't as balanced as the numbers suggest also lead to a subpar experience at the table. It's an invaluable guide that most people will miss the minute they plan a combat in a different game.