To be fair, though, the lightsail instances that are subject to oppressive throttling aren't really a good comparison. Those should be dirt cheap, as they are useless if you use them in almost any real fashion.
I had some personal projects a while ago that I used to run on Linode, I later moved them to LightSail and I never noticed any performance difference in these projects running there.
It depends on the instance. If you use up the burst capacity, you will notice. The very bottom end only allows about 5% usage before the throttling kicks in. There's a chart here:
Hmm, interesting. I unfortunately don't have the metrics anymore since this project is long dead.
But I do remember on another host (I tried a different one between linode and light sail) that I actually got an email that I was using too much CPU and was shut down.
So I do remember using a lot of CPU, but I don't know where that percent was.
Doesn't LightSail rely on t-family (burstable) instances? If so, that's probably because your projects' regular sustained usage do not go above the minimum threshold (which is fairly low).
These were Ruby on Rails projects.