The question itself is I would say indicative of a deeper problem here, and that is a mismatch between different expectations the players have of the GM (or of the game in general). Most likely, there are two statements applying to your gaming group at the same time:
- The players don't want their characters to die
- The expected outcome of every combat scene is complete annihilation of one side
The problem is, quite simply, that those two things can't be true at once without making combat completely pointless. Trying to find some "compromise" between the two can be really hard for GMs, especially if they don't have much experience. Therefore, my advice is less mechanical and more general:
Do not consider combat a Players vs. GM situation
It's an easy trap to fall into because of the wargame-like nature of combat rules, to turn the roleplaying game into a board game between the GM and the players which has a clear "winner". This just creates unnecessary bad feelings, because once you have the expectation that somebody has to loose in a situation, you don't want that person to be you.
But a GM doesn't "win" by killing off player characters. This is true in all situations, even in combat. Putting aside the mindset of "my monsters have to succeed against the party" will help you approach the problem more relaxed.
The win state of a GM, if one can talk about such a thing, is to provide an engaging game, and that means providing people with situations and choices to roleplay. Combat which is just about "kill the other side" doesn't provide much choice. Therefore, this leads right into the next point:
Make combat have different end states than extermination
Think for a moment about why a fight happens. What do both sides want out of it? Will a group of henchmen really keep shooting at the heroes when they effortlessly pick them off one by one? Why do both sides consider employing lethal force as the right tool in this moment?
Once you know the answers to that, you will find ways for a fight to end without either side completely eradicated. If the reason for a fight is no longer valid, or the chances of success too small, people will stop fighting. Have enemies surrender, or try to run away, or stop once they have sufficiently damaged the party to "send a message" and so on. This way, combat turns from a simple "slaughter the other side" exercise into a roleplaying opportunity. What do the player characters do when the enemies thrown down their weapons and plea for their lives? What do they do when the group disengages after having knocked out one character and having dragged them off the battlefield?
And once you have internalised that, then you can get to the mechanical solution:
Switch up tactics
Right now, your game is similar to Rock Paper Scissors where you always play Paper and they play Scissors, with the obvious results. And your question turns into "How do I beat Scissors without making my opponent loose?", which is unanswerable for obvious reasons. Adapting the above mindset to combat allows you to play Rock without immediately feeling bad, and turns the question into a solvable one of "How do I play Rock?"
And there might be an answer to that, but in case nobody on this site knows it, having re-framed combat to be not automatically lethal allows you to use your player's expertise to your advantage: Let the enemies focus fire on them, and then see what tactics they come up with to counter it. Then adapt those. This sounds like an arms race, but the point here is that you're not doing this to kill the party, but to allow the mobs the successes they need to carry out whatever not-kill-everyone goal they have going into combat.
And at some point, the arms race turns into Rock Paper Scissors: The new strategy to beat the old counter-strategy is one already known. At that point, your combat will have been greatly enriched, not only by the above inclusion of roleplay, but because the decision of "what strategy will we try today?" is inherently more interesting than the constant repetition of one winning strategy.
