I'm so glad you asked
1) Picking a commander
In years past, I went through every possible commander in the format and gave them all grades based on how interested I was in building them. This was back when the number was <1000, so it wasn't quite as insane as it would be now. I considered each of them, how one might build them, and then gave them a grade in a big spreadsheet that took me ages to make. For a while, that was my main source of inspiration - look at the top picks until one spoke to me.
These days, it's pretty rare that a I brew an older commander - typically I get interested in a new commander during spoiler season, and then use my GBU article as an opportunity to gather my thoughts on each of them. This is similar to my old list, except that I'm a lot more thorough since a) there aren't nearly as many options to consider and b) publishing it out to the world means I'll feel like a moron if I miss a compelling element of the card.
But sometimes I will get a hankering to build something older, or by happenstance some older commander will get my renewed attention - I notice it while combing through my collection, or see it in a trade binder, or see a decklist online.
One thing that increasingly effects my interest in a commander these days is how popular it is on EDHrec. A commander that I thought looked cool get tons of decklists can dramatically lower my interest in it, and a commander getting underused can be absolute catnip for my flighty inspiration. Good example recently was
Gilraen, Dúnedain Protector. Wouldn't have given her a second thought if she had a few hundred decklists even, but seeing her almost completely ignored (still only 45 decks!) made me think "now hang on, she's not that bad, you could do X and Y and..." and suddenly my gears are turning.
2) Picking a strategy
This typically goes hand-in-hand with choosing a commander. Rarely do I pick a commander but not know where I plan to go with it. Nevertheless, it's still an important step in the process. Much like my contrarian commander selection, I really like finding unique ways to use a commander. For example, on another thread,
Rukarumel, Biologist came up, and based on decklists a significant majority of people are basically building her as sliver tribal, which I think is a poor utilization of card and not very interesting besides. Seeing that Rebel "tribal" lists are a significant minority for Ruka is definitely something that makes me more interested.
As far as how I decide on the strategy, it's difficult to communicate since a lot of it is just experience, but beyond obvious synergies I think I'm fairly good at weighing many different possible game plans and synergies against a given commander. For examples of that, I suppose just look through my GBU articles where I try to consider many different gameplans for each commander, beyond those that might be immediately apparent. Or of course, the crown jewel, my
Phelddagrif list, which (at least at the time) was completely different from how anyone else I'm aware of brewed the purple hippo.
3) Gathering cards
I've mentioned many times on here that, besides a few "permanent decks" which are normal, self-contained decks, I keep a collection of every "commander-viable" card (in my estimation), all sleeved and sorted (by color and land/nonland - I don't see much point to sorting beyond that). Currently that collection numbers 11,830 and is worth $62,086.36, according to the app that I use to track it (the "Decked" app, if anyone is curious).
Once I've picked out a commander and have a rough idea of how I plan to use it, I start going through the nonland cards within its color identity, typically starting with multicolor and working my way to colorless. Along the way I'll find cards I had intended to use, cards of types I expected to use, and sometimes cards that hadn't occurred to me but make a lot of sense once I'm looking at them.
I can filter through physical cards pretty quickly. I've tried to do the same thing with scryfall and other online resources and it's never been nearly as quick. Physical cards also make the following stages significantly easier for me.
4) Sorting
With a better understanding of what the deck will be doing, now that I've seen all the cards that fit into my planned strategy plus anything else I might not have considered, I start to sort the cards into the different functions I consider to contribute to this strategy. For example, working on a
The Rani deck, in contrast to how it's typically built, I'm eschewing other forms of goad and instead relying on my commander to provide a sufficient amount of goad. So the piles for that deck include things like board control tools to force opponents down to a smaller number of creatures (or at least a small number of attackers) to ensure that I have enough goad to cover the important enemy creatures, artifact synergies that trigger off number of artifacts to exploit the clue tokens she creates, artifact token production to further increase the value of those synergy pieces, ordinary targeted removal, counterspells, and card advantage.
5) Culling
Each step of this process further helps me understand what the deck is going to do, so now that I've sorted out all the piles I'm starting to form an idea of how important each function will be. Is this something I only need occasionally? Or is it something that I absolutely need to have in the first couple turns?
So now I go through each pile and start to remove the weaker options. Stuff that seemed okay at first is often now clearly outmatched by other cards, and can be culled.
Not every time, but especially if I'm taking the deck pretty seriously, at some point in this step I'll create a blueprint for the deck that will have a number of cards attached to each function. Once I have those numbers adding to 99, in principle I can cut down to the top X cards for each category and bam, list complete. Of course in practice this is a bit messier because some cards will be part of multiple categories, and some cards are reliant on other synergy pieces. So it's an iterative process of cutting, reconsidering, perhaps un-cutting some cards, adjusting the numbers on my blueprint, and then cutting some more.
6) Lands
Lands are almost always the last thing I add to a deck, once I know the balance of colors and have decided how land-heavy or land-light I want the deck to be. This process isn't always so different between decks - the best lands are usually still the best lands, though sometimes certain features of the deck will have an impact, i.e. multi-pipped cards increase the value of filter lands, slower decks can more readily accept ETBT lands, mono-color decks can accept more colorless utility lands, etc. And sometimes utility lands can be an important part of the deck as well.
7) Posting
Once the list is complete, I make a new thread and stick the decklist on there, alongside some thoughts I have about it. A new handmade Dirk original is complete, and the cycle begins again.
8) Actually playing the deck
Optional