Four hours outside of Phandelver the party ran into an ambush set by four goblins.
The party roll perception rolls. The goblins rolled various 20s for their stealth check. The Goblins got a surprise round.
In the surprise round, the goblins shot arrows taking out both the human wizard and the elven wizard.
The Goblins win initiative over everybody except for the downed wizards. They shout a NPC Wagon Driver and the Human Fighter who remain standing.
The human rogue starts running towards the goblin shooting his short bow. The human fighter dashes toward the nearest goblin.
The Human Wizard rolls a natural 20 on his death check. The Elven Wizards get a successful death check.
The next round the goblins focuses on the charging Human Fighter but his high armor class prevents him from being hit.
The Human Rogue closes in and kills a goblin with his short bow. The Human Fighter reaches a goblin. The Human Wizard hides. The Elven Wizards continue to roll death checks.
The next round the Human Wizard cast sleep causing one more goblin to fall.
The remaining two goblins start running away
The Human Rogue shot down one goblin, and the Human Fighter kill the last goblin.
The fight is over with all goblins down. The Elven Wizard is stabilized.
Surprise is important and goblins are good at creating a surprise round due to their high stealth.
In general low CR 5e monsters have one special ability they are good at. This can be decisive under the right circumstances.
Quantity is also a decisive advantage. For another group with 8 PCs I ran this encounter with 8 goblins. The goblins were completely outclassed even with surprise. It is my opinion that the multiplier for number of opponents needs to be used for the party size as well. In subsequent session it is obvious that doubling the monster does not provide the same challenge if you double of the number of PCs. It wasn't until I increase the difficultly to four times the original I was able to get comparable results for the eight PC group as I did for the four PC group.
5e combat is highly situational. Different plans, different terrains, different initial conditions can produce widely varying results. The result is that small differences in CR don't mean much. Only when the numbers are increased from 50% or 100% on either side the differences become decisive.
5e rewards system mastery but there is less to master. And because of 5e combat sensitivity to circumstances, there is no combinations of abilities that make for an instant win.
The use of a d20 and the flat probability curve means that a run of bad or inferior dice rolls can and will happen. The same with a run of superior dice roll. In combination with 5e's sensitivity to situational factors this means results can vary wildly from group to group even when using the same PCs.
In general the book values work great for four man parties. Try running a few encounters with a four man party, Phandelver is good for this. Do this to get a feel of how 5e combat is supposed to be like. Then for a larger group, increase your encounter size by 25% increments until you get the same feel as the smaller group.