Depends on your development and per-action cost. And on the possible latency. It also changes your whole stack from "send a request" to "emulate each step in a browser while taking screenshots at (hopefully) the right event/delay" - that's a huge difference.