I find the guaranteed nature of the effects matter because of how much they contrast with skills in terms of flat D20 rolls.
I see this in player planning, and in the kinds of plans players instinctively don't make in comparison to other games like Savage Worlds where the dice have more of a bell curve and there are ways to mitigate failure with metacurrency.
In D&D a plan that involves multiple steps involving skills rolls is a bad plan, "I'm going to climb the wall, sneak across the courtyard, knock out one of the cultists, steal their robes and then infiltrate the inner sanctum during the cult leaders sermon" is not a good plan if the DM is going to ask you to roll for each of those things. It's bad to the extent that players will probably not put it forward in those terms, so it's easy to miss the issues.*
"I'm going to spider climb up the wall, cast disguise self so I look like a cultist and walk into the inner sanctum (and then Misty Step out if things look dicy)" is a workable plan. You're guaranteed to at least the do the necessary thing to get to next stage. At some point you may need to roll a skill, but the less rolls the less possible points of failure, the more workable the plan.
Magic, in terms, of spells, and magic items have always been the main problem solving devices of D&D. If the players have access to these they'll come up with something. If they're relying on skills...well you'd better be prepared for a lot of failing forwards.
*There are ways to mitigate this of course, by either rolling once for the entire plan (pretty unsatisfying if you fail though) using some kind of skill challenge (not present in 5e).