So, when should you actually put in a load balancer?

hey everyone, im trying to get a better handle on when it makes sense to put a load balancer in front of our services. were seeing increased traffic on a few key apis, and im wondering about the tipping point. is it purely based on server cpu usage, or are there other metrics like request latency, error rates, or just the sheer number of instances were running that should signal its time? what are your thoughts on the common indicators and perhaps some simple rules of thumb?