Ok.
I'll tell how I would do it.
First things first, no two GPUs will overclock the same. Even GPUs that are the exact same make/model will have some discrepancies, therefore using clock settings from someone else's overclock is something I would advise against. It's much more prudent to find the limits yourself.
First thing I would do is stress test the card at default clocks/voltage/everything using a benchmark with everything maxed. Doing that you can get a base benchmark to compare to and you have a little assurance that you didn't get shipped a defective card. If it fails (BSOD/black screen/artifacts) I would recommend contacting the manufacturer. Sometimes (rarely) a card can be shipped with improperly applied thermal interface material. Either way in all likelyhood you will pass this.
You will notice I use the word artifacts quite a bit. I'm not the best at defining what an artifact is, but generally it is anything that looks out of the ordinary when the GPU is rendering to your screen. Little white dots, stretched textures, lines, it could be alot of things. If something looks pretty wacky, assume it to be an artifact.
I would get a pen and paper to keep track of things, makes things a little simpler.
Check what the max operating temperature and voltage is for your GPU. Write this down. Google is your friend here, try to use reputable websites, manufacturers websites are usually reliable.
Watch your temperatures constantly. Turn the fan up ideally using a custom profile that let's you set fan speed vs. gpu temp if your GPU is getting hot. Try to keep the fan under 75C, I would advise avoiding going over 80C. Ideally you want to stay under 70. The max for most GPUs is around 90C. Again, do your research and don't rely on 90C being the max, try to get specific information for your GPU.
Write down the default core clock and memory clock of your gpu. Next thing I would do is raise the core clock 5-20 mhz (depending on how patient you are really, starting off with a higher increment than using 5 later is an option as well). Stress test the card and look for artifacts. If you BSOD or black screen you've hit the limit. Repeat this step until you hit the limit. When you hit the limit lower the core clock by whatever amount you last raised it, then stress test. If you pass the stress test, write down the core clock. Keep on eye on whatever "scoring" the benchmark gives you to be certain that the card is actually performing better at the higher clock frequency.
Now set your core clock back to the default.
Now what I would do is set the memory frequency 5-20mhz higher. Stress test using whatever benchmark you chose. Repeat this step until you hit the limit. When you hit the limit lower the memory clock by whatever amount you last raised it, then stress test. If you pass the stress test, write down the memory clock. Keep on eye on whatever "scoring" the benchmark gives you to be certain that the card is actually performing better at the higher clock frequency.
Now you should have both a memory and core clock down that we assume to be stable. Set the GPU to these clocks and stress test. IF it fails (artifacts/BSOD/BS) back the memory clock down until it passes.
Now that we have what we assume to be a stable overclock, I would recommend running OCCT for 3 hours. If you are impatient or 3 hours is just impractical, run it for 30 minutes. Watch your temperatures carefully when running this test.
http://www.ocbase.com/
If you want to get higher clocks by increasing voltage, I won't be the one to help you with that
I'm not comfortable guiding you threw that since if I tell you something wrong there's a much higher risk of damaging your GPU. Maybe someone else can help you here if you are dead set on getting the absolute most you can out of that GPU. Keep in mind that very small increases in voltage make a big difference, both in stability and heat.
Some common benchmarks used to stress test GPUs:
Furmark
Unigine Valley (or Heaven)
Facebook
Twitter
Instagram