Long answer is also practice, but with information lol.
I spent a long time moving from 2 look to 1 look (oll and pll) using this website jperm.net/algs/pll I’ve linked the PLL but there’s OLL there too.
On this page I learned as many of these as I could. You can click on the picture and change the status from unlearned to learning to learned, then go to the trainer and you can select those 3 statuses and it will show only things you’ve selected. You can also click on the alg to get alternatives or even put your own alg in.
So I’ve gone to that page and set an alg as training, then just practiced that one alg until I have it in muscle memory. I aimed at trying to get each type of alg learned, so corners are there, or middles need moving, or headlights are there, or even that there is no pattern.
I was advised to learn PLL first then OLL algorithms, but I kinda picked out the patterns I see most in both, or algs that seemed the simplest first. So I have been learning both together and working from easiest to remember to hardest, and also making my own up for the ones that are most complicated.
This meant that I could practice a few, then go and do solves and when the patterns I knew came up I’d get faster solves for those patterns.
Ok so that’s the big time saving out of the way, takes a lot to learn all the algorithms so it’ll take time. But there’s also taking time to plan your cross. That can save you a good chunk because there’s less head scratching when you start.
Then there’s the look ahead, which I’m only just getting. I did a lot of slow solves to get this in my brain and it’s quite big. This is what I’m practicing to get from my 40second average down to 25
So as you’re solving a corner and edge into the corner, once you have it set up into a 3 move insert, you don’t need to look at it anymore. It’s 3 loves to insert, so instead of looking at it as you put it in, you have to train yourself to look for the next 2 pieces you’re putting together.
While your learning the OLL and PLL and just doing solves (not training algs), when you get one you don’t know, try and alg you do know on it. Sometimes this changes your top pattern to an easier pattern that you can solve. It’s like a stopgap 2 look (pll or oll). Eventually as you learn all the algs you’ll find that you can “wing it” with some of the harder algorithms by just doing a couple of easier ones, which is all the harder ones are anyway, a couple of easier algs with a connecting move in the middle.
Hope all that helps. I’d also advise you have a “travel cube.”
I have an extra cube that lives in the pocket of my leather jacket (yes I’m a metal-head Dad, doesn’t quite fit the stereotype does it?) and it comes in useful when I’m stuck in a queue, or at the Dr waiting room or A and E (or emergency care as it’s called in the US). This allows me to cube instead of whipping my phone out when I have time to kill out of the house.