The turning point for most with lookahead is when you start to understand the principle of it.
Imagine a CFOP solve
You find your cross solution during inspection.
You are most likely wired to look at your solution while performing it, making sure you are doing everything right.
Now you have a pause for your first pair
No lookahead
looking ahead means that you don't look at the cross solution you are currently doing, but instead look at the pieces required for your first pair.
So when you are done with cross, you know where your two pieces are so you just need to recognise the case(or eventually predict it, eliminating the pause entirely)
lookahead
This is just an example for your first pair.
You do the same thing for all the other 3 pairs. Once you recognise the case, let muscle memory take over and find your next target. Eventually, this opens up things like predicting cases beforehand, EO during F2L.
How do you start to do that?
Start by turning real slow. Maybe not 1 TPS, because that get's boring quickly and you'll end up speeding up by the second solve. Somewhere between 2 and 3 TPS is a better start. Take that turning speed constant and start looking ahead.
If you still make pauses, drop the speed, if you start to get the hang of it, increase speed.
Decreasing your turning speed until you can lookahead with your normal speed will help you in the long run.
Same thing of course applies to any other method, difference being that writing this text CFOP-centered leads to the least angry comments.