Part of what helped it be successful was that I was, for over a year, an active commentator and critiquer. I ended up internalizing many aspects of what people pointed out as a good or bad SCP, and kept coming back to the lowest rated pages, seeing how and where articles stumbled.
I also made it a point to read all of the guides. Not in one sitting, mind you, but over time.
During that time, I wrote up drafts for other things, and getting them reviewed on the forums, but never submitted them. They were lagely practice in the format. Keep this in mind: making a Successful First SCP is not the same as your First SCP Idea. Knowing what Ideas are good to start with, and which ones you have a good chance at pulling off is key. You won't know this at first, but if you join in discussions and practice the format (while getting feedback), you'll get a better idea of what you can do.
I didn't work much on the idea until I heard that the Short Works Contest was a thing. I took the basic concept, "knife what cuts through dimensions", and ruminated on what it could mean. In this case, "dimension" could easily mean "alternate reality" and "cut through" more-or-leas means "severed". Hence, a knife that severs alternate realities, and took that to it's logical conclusion (destroying a universe).
Once I had the idea ready, I moved it to the Drafts forum, and then Chat (both #site19 and #site17). In Chat, I listened to the feedback, with most reviewers expressing interest in what happened during the testing phase, or giving a generic "It's missing something" reviews. Keep in mind, at the time I didn't have the collapsed section on 2207 that is there now, so as to meet the wordcount limit for the Short Works Contest.
So, I shifted gears and expanded the idea to include logs. This is another tip: you should listen to feedback if multiple people tell you to change or improve something, even if doing so significantly alters what you had planned. It's easier to work with the grain than against it, and trying to go down a route that multiple people are indicating that won't work will more likely fail than suceed.
After adding the logs, I once again went to chat for reviews, as many as I could get for about a week or two. Once multiple people said that they couldn't find anything wrong or anything to modify, I posted. This is an important tip: don't post until the majority of people over a period of time say that they can't find anything wrong with it. One person's opinion won't match with everyone else; only a poll of multiple people can show how well an article is posting.
Only then did I post, after using Microsoft Word to replace my number placeholder for the actual number.
(account deleted)