You can already solve the 3x3 layer by layer, it takes you a minute or two each time, and now you want to go faster. This post is a step-by-step plan for how to solve a Rubik's Cube faster: what to practise at each time goal, how to measure your progress, and which techniques really help you get under 30 seconds.
Find your case fast
Look at your cube from above and tap the picture that matches it to see how to solve it. Grey cells are stickers you can ignore.
Good news first: you do not need to know hundreds of algorithms. J Perm, one of the best-known cubing YouTubers, writes that some people know lots of algorithms and still take over a minute, while others solve in under 30 seconds with just the beginner method. Practice matters more than how many algorithms you know (J Perm).
Every move sequence in this post was tested on the EasyRubic cube simulator.
First: time yourself the right way
To get faster, you need to know where you are now. A single solve tells you very little: sometimes you get an easy scramble, sometimes a hard one, sometimes your hands shake. That is why cubers use an average.
What is ao5?
Ao5 (average of 5) is the average of 5 solves, but you drop the fastest and the slowest one and take the average of the 3 that are left. This is also how the official 3x3 result is worked out at WCA competitions: each person gets 5 attempts, the best and worst are dropped, and the middle 3 are averaged (WCA Regulations, 9f8).
For example, say your 5 solves are:
| Solve | Time |
|---|---|
| 1 | 32.15 seconds |
| 2 | 28.40 seconds (fastest, dropped) |
| 3 | 35.02 seconds |
| 4 | 29.87 seconds |
| 5 | 41.30 seconds (slowest, dropped) |
Ao5 = (32.15 + 35.02 + 29.87) / 3 = 32.35 seconds. WCA rules round averages to the hundredth of a second (WCA Regulations, 9f1).
If one solve is a DNF (did not finish), it counts as the worst solve and gets dropped. With two or more DNFs, the whole ao5 is a DNF (WCA Regulations, 9f9).
Ao12 and longer averages
According to the SpeedSolving Wiki, an average of many solves drops the fastest 5% and the slowest 5% of the solves, rounded up (SpeedSolving Wiki). In practice:
- Ao12: drop the 1 fastest and the 1 slowest solve, then average the other 10.
- Ao50: drop 3 solves at each end.
- Ao100: drop 5 solves at each end.
Our tip: use ao12 as your main measure when you practise at home. It is long enough that one lucky solve will not fool you, but short enough to finish in one practice session. When this post says "under 30 seconds", it means an ao12 under 30 seconds, not one lucky solve.
The EasyRubic 3x3 timer page gives you scrambles and times each solve.
The plan, one time goal at a time
The goals below are how we split the road for beginners, not anyone's official rules. Each goal has one main thing to practise. Once you have that down, your times usually drop on their own.
Over 2 minutes: solve without looking at the guide
If you still need to open the guide halfway through, that is where you lose the most time. Go back over the 3x3 guide until you can solve the whole cube without stopping to remember the next step.
The step that makes beginners pause most is putting the layer 2 (the middle layer) edges in place:
- You start with (input)First layer all correct
- Result you will get after finishing this stepFirst two layers all correct
Goal for this stage: 5 solves in a row without looking at anything. Do not worry about speed yet.
From 1 to 2 minutes: finger tricks and fewer pauses
By now you remember all the steps. Most of your time goes in two places: slow hands, and the pauses while you look for the next piece.
For your hands, get the hang of finger tricks: turn with your fingertips instead of your whole hand. The first drill is R U R' U', done smoothly without letting go of the cube. Do it six times in a row and a solved cube comes back to solved (checked on the simulator), so you can check for yourself whether you made a mistake.
- You start with (input)
- After the formula
For the pauses: after each practice session, think back to which step you paused at the longest. Usually it is finding a white corner (step 2) or a layer 2 edge (step 3). The "Look-ahead" section below explains how to pause less.
From 1 minute to 45 seconds: cross on the bottom
The beginner guide makes the white cross on the top face, then flips the cube over. Fast cubers build the cross straight on the bottom face. The SpeedSolving Wiki explains that most cubers recommend doing the cross on the bottom, because you can see the pieces you need for the next step more clearly, and you do not waste time flipping the cube (SpeedSolving Wiki).
It will feel hard at first, because you cannot see the white face. The trick is to look at the white edges from the side faces. Here are the two easiest cases, with white on the bottom:
The edge is in the top layer, right above the centre of the same colour, with its white sticker facing up. Turn that face twice: F2.
See it in 3D, before and after
- You start with (input)
- After the formula
The edge is in the top layer, above the front face, but its side colour matches the right face. Move it over to the right first, then turn the right face twice: U' R2.
See it in 3D, before and after
- You start with (input)
- After the formula
The U' move matters: it brings the edge right above its spot before you turn it down, so it does not mess up the white pieces already on the bottom.
According to the SpeedSolving Wiki, the cross can always be done in 8 moves or fewer (SpeedSolving Wiki). J Perm writes that with lots of practice the cross can take under 5 seconds (J Perm). Every case has pictures in the post How to do the CFOP cross.
From 45 to 30 seconds: a shorter last layer, then F2L
At this point the beginner method starts to hold you back: layer 3 (the top layer) takes four steps (steps 4 to 7), and many algorithms have to be repeated several times.
J Perm recommends tackling the last layer first, then F2L, and leaving cross improvements for later (J Perm):
- 2-look OLL: make the yellow face in at most two steps, with 3 algorithms for the edges and 7 for the corners. See the post 2-look OLL.
- 2-look PLL: put the last layer pieces in their right places in at most two steps, with 2 algorithms for the corners and 4 for the edges. See the post 2-look PLL.
- Intuitive F2L: pair a corner with an edge and put them in together, instead of steps 2 and 3 of the guide. J Perm estimates this saves about 20 moves per solve. See the post Intuitive F2L.
Two algorithms you already know from the guide are also part of 2-look OLL and 2-look PLL: Sune (step 5) and the algorithm that swaps three yellow edges (step 7). So before picking up any new algorithms, get these two really fast.
Sune, yellow face on top: R U R' U R U2 R'
See it in 3D, before and after
- You start with (input)
- After the formula
Swap three yellow edges, yellow face on top: R U' R U R U R U' R' U' R2
See it in 3D, before and after
- You start with (input)
- After the formula
Drill: do Sune six times in a row on a solved cube and it comes back to solved. For the edge algorithm, three times in a row brings it back. Both were checked on the simulator. Time each round, then try to make the next one faster without mistakes.
Do not take on everything at once. J Perm warns that cramming too many algorithms at once makes you forget them (J Perm). Pick up one or two algorithms, use them in real solves for a few days, then move on to the next.
And do not give up when you switch to F2L: J Perm writes that it may feel slower at first, but it gets much faster once you are used to it.
Four moves.
One at a time.
Tap each symbol and follow one piece. This is an exercise in watching the movement.
Look-ahead: spotting the next piece early
The SpeedSolving Wiki calls look-ahead "the art of seeing what is coming": while you are doing one step, you are already finding the pieces you need for the next one. Cubers with good look-ahead solve with almost no pauses, and that is why they are fast (SpeedSolving Wiki).
Here is an example with step 3 of the guide. You are doing the algorithm that puts a layer 2 edge in on the right, with the yellow face on top:
U R U' R' U' F' U F
See it in 3D, before and after
- You start with (input)
- After the formula
You already know this algorithm by heart, so your hands can do it without thinking. So while your hands work, do not watch the edge you are putting in. Look at the top layer and find the next edge without yellow on it. When the algorithm ends, you already know how many U turns to do and whether to use the left or the right algorithm.
The simplest way to practise: solve more slowly. Solve the whole cube at a speed where your hands never have to stop, however slow that is. Your times will be worse at first. But once you are used to looking ahead, you can speed up bit by bit and still keep a smooth, pause-free rhythm.
A sample 20-minute practice session
Here is the session we suggest, so you have a clear plan instead of just solving over and over:
| Time | What to do |
|---|---|
| 3 minutes | Hand warm-up: R U R' U' and Sune, a few rounds of six each |
| 5 minutes | Practise your slowest step on its own (for example: only do the cross on the bottom, then scramble again) |
| 5 minutes | Slow, pause-free solves to practise look-ahead |
| 7 minutes | Normal timed solves, get one ao12 |
Pictures for the two warm-ups:
- You start with (input)
- After the formula
See it in 3D, before and after
- You start with (input)
- After the formula
Write down your ao12 every day. Look at the week, not each day: some days you will be slower than the day before, and that is normal.
The cube and everything outside your hands
- A good cube really helps. A stiff cube that catches a lot will ruin your finger tricks. If your cube is old, see the post Which Rubik's Cube to buy as a beginner.
- Inspect before you solve. At WCA competitions you get under 15 seconds to inspect the cube before you start. Go over 15 seconds and you get a 2-second penalty; go over 17 seconds and that attempt is a DNF (WCA Regulations, A3a1 and A4d). When you practise at home, use that time to plan your cross.
- Competing. Once your ao12 is steady, going to a WCA competition is a great way to get an official result. See the post WCA Rubik's Cube competitions.
Mistakes that make you stall
- Solving over and over without knowing which step is slow. Every now and then, time each step on its own.
- Picking up new algorithms too early. If you still have to think about the old ones, new ones just add more places to pause.
- Going all-out on every solve. When your hands are faster than your eyes, you get long pauses. Mix in some slow solves.
- Looking at single solves instead of the average. One 25-second solve does not mean you are sub-30. Only your ao12 can tell you that.
To understand the fast method this post leads to, read What is CFOP.
Sources
Frequently asked questions
Can you solve a Rubik's Cube in under 30 seconds with the beginner method?
Yes. J Perm writes that some people solve in under 30 seconds with just the beginner method. But the long last-layer steps of this method make it harder, so picking up 2-look OLL and 2-look PLL is usually the quicker road.
How do you work out ao5?
Take 5 solves, drop the fastest and the slowest, then average the 3 that are left. This is also how 3x3 results are worked out at WCA competitions.
Should I start with F2L or with OLL and PLL?
J Perm recommends starting with OLL and PLL (the 2-look versions), then F2L. The last layer gets you faster results with fewer algorithms, while F2L takes time to get used to.
How long does it take to get under 30 seconds?
There is no single answer, because it depends on how much you practise and how. Keep track of your ao12 every day: when your times stop dropping, that is a sign to move on to the next goal in the plan.
What is look-ahead?
It is the skill of finding the pieces for the next step while your hands are still doing the current one, so you do not have to stop between steps. The easiest way to practise is to solve slowly enough that your hands never stop.
This information is based on competition results owned and maintained by the World Cube Association, published at worldcubeassociation.org/results as of 12.10.2026.
Cover photo: Explodingcreepsr / Wikimedia Commons, CC BY-SA 4.0


