...As far as you know.
So here's my method-related confession: I often start new conlangs just to try out new methods of building them. That is the ultimate origin of both ea-luna and Teliya Nevashi. In each case, I started out with a few goals related to the language itself and the method of creating it. Both languages were experiments, not intended to be realistic nor to have the sketchy fictional cultures that my imagination later attached to them.
ea-luna was meant to be more organized and systematic in the way it was created than my previous efforts, while being as different as possible from those earlier language sketches and hiding the fact that the base vocabulary was built in a systematic way. I made it appear more random than it actually is by assigning words from an old version of the Universal Language Dictionary (plus other words I felt were necessary or culturally relevant) to a handwritten list of all possible one- and two- syllable words, moving across the columns instead of down (with four columns to every two page spread). In retrospect, listing all those words ea-luna-to-English in ea-luna syllabary order may not have been the wisest choice for looking things up later. There are a few elements that work as affixes for derivation, but, for the most part, new words are compounds of the words in the original notebook, which was all well and good except that I kept that list in a different notebook that was lost in one of our many moves. The lack of documentation of ea-luna grammar is a nightmare all these years later-- what remains are unlabeled and sometimes untranslated paradigms, examples of sentences of different levels of complexity, and cryptic notes to myself.
I scrapped that method of word generation after ea-luna, although I do see value in a basic "must have" list of concepts to be adapted to the project. There's value in knowing what kinds of words are possible and having a bunch of them sitting around at any given time, just waiting to be assigned a meaning, as a convenience. I see some merit in building through use/translation, but if I were using that method now, I'd describe all the new emerging features as they emerged, rather than trying to figure out the rules I used later.
I began work on Teliya Nevashi with 3 things in mind: I wanted to get used to using the computer in my conlanging (specifically so that I could share my projects online more easily), to experiment with verbs, and to build a word-by-word crafted language. Nevashi was started with a barebones grammar with a few dummy words first, and then I built words one at a time as needed (or derived them from existing words). I have an idea of what Nevashi words should sound like, but the rules for arriving at that are incomplete at best. This is at least in part because everything that starts with phon- is horribly boring to me.
It turns out that creating words one at a time as needed is also tedious, and creating whole lists of words at once is right out the window; I always end up with a bunch of words that sound too similar to one another when I create a lot of words together. Using the computer turned out to have a number of advantages aside from making the language easier to share: there's the searchability that is lacking in handwritten documents, I can easily make a second copy so that I can make changes without losing what I've done before, and because I save my work on Google Drive (formerly Google Docs), it is accessible from any computer on the Net.
I mention all of this for several reasons. First, I intend to try out some new methods soon, and I have a few different language ideas floating around in my head. I'd like to work on a naturalistic/realistic, diachronic, a priori language with a whole cultural background n'at*, and an a posteriori could-be-an-auxlang-if-it-weren't-so-lazy kind of conlang. I've got some long overdue collaborations that I need to get down to also. Looking at where I have been gives me some idea of where I ought to go. Or at least where I might go. Secondly, I am thinking about interviewing people about their methods, and it's only fair to show mine before asking anyone else to show theirs. Lastly, this is my answer to why I haven't collaborated on anything with Jeff Burke (there at LCC2, and here at LCC3) despite the fact that we talk to each other all the time. We share a lot of the same madness, but none of the same methods.
*I live in Johnstown, PA. I knew I'd started to go native the first time I heard myself say "n'at" ("and that"). I haven't started saying "yinz" though. I am from the land between "yous" and "y'all", and I still fall on the "y'all" side of the fence.