It's the same concept as, "Why bother playing out a disagreement between the PCs and NPC(s) if there's no risk of death". Conflicts are likely to arise numerous times during a campaign, and it's only the most basic of hack and slash dungeon crawls where death is always a risk.
<snip>
I see a campaign where you can't die as working similarly.
<snip>
In such a campaign, I expect that combat is simply another approach to conflict resolution.
<snip>
It depends on the group in question, but for some groups in-game consequences are far more important than stats.
This is a pretty good description of a "non-risk-of-death" approach to play.
My 4e game is by-the-book with only some very minor tweaks around PC build options. So the risk of death is the mechanical default for 4e ie there, but not very great. In the course of a campaign from 1st to 28th level with 5 PCs, there has been one "TPK" at 3rd level, and four other PC deaths.
With the "TPK" - at the hands of dead spirits conjured by a goblin shaman - I asked the players whether or not they wanted to continue with their existing PCs. All but one did, and only one of the PCs had actually died on stage, being dropped below negative bloodied hp by "friendly fire".
So 3 PCs recovered consciousness in the goblin prison, with a 4th prisoner there (the new PC for the player who wanted to change characters). The PC who had died was returned to life by his god (the Raven Queen) on a ceremonial slab where the goblin shaman was using his body as a focus to summon another dead spirit to whom that PC had had a strong connection.
Of the other four deaths, in 3 cases the PCs arranged for Raise Dead to be cast. The first happened at 2nd level, so Raise Dead was beyond their abilities, but again the player wanted to keep playing the PC, so he was sent back from death by the Raven Queen to recover an artefact hidden in the ruined Nerathi site where he had died. The idea of an artefact as the reason for being sent back was invented by the player (as best I recall) - it ended up being the first part of the Rod of Seven Parts, which has been a major focus for that PC and the campaign in the 25 subsequent levels.
As well as 4e, I've recently been running a bit of Burning Wheel. The threat of death in that system is also pretty low, but the players nevertheless care about failure because things can go wrong (in the fiction). For instance, in our most recent session the ship the PCs were on sank because they were unable to rally the crew to put out a fire in the rigging. The PCs were then rescued from the sea by a passing Elven ship (the player of an Elven princess made a successful Circles check, which is the BW mechanic for meeting NPCs from out of your backstory). The captain of the ship put two of the PCs in the brig, because they were marked by the Shadow (the two PC sorcerers), and put them off the ship on the shore of the Bright Desert. The Princess jumped ship with them, because she was concerned that the captain was going to make her return to the Elven court, whereas she was trying to make her way among humans. So now the PCs have to make their way through the desert.
For players who are following the fiction, fictional setbacks are the consequences that matter. Death is only one of many possible fictional setbacks.