Without digging up my own notes.
I agree that rolling is important, the biggest hurdle I struggle with is three-fold.
1) Time
2) Wealth
3) Realism
To #3, it bothers me a lot that you can make multiple gallons of Ale in a single day. I know only a very little about brewing from some research for a character, but that is not even close to something drinkable as Ale as far as I've been able to tell.
Which leads to an annoying issue that each craft or item would have to have a different time scale. I did find a site where the author went into this, but I got sidetracked before delving deep enough to see if I could take it whole cloth. Link here
D&D 5e – Questionable Arcana – Mundane and Magical Makings: An Alternative Crafting System for 5th Edition
One thing I did want to take from it was an alteration of the training rules. Essentially, you could gain increasing bonuses by the GP value of your work. So, make 10,000 gp worth of blacksmithing, and you gain expertise (numbers pulled out of the air just to demonstrate the idea) the part of it was, I did the math, and the numbers were intense. I think it worked out to be close to 3 years of work, which makes sense. Mastering a craft is a skill that takes immense time and dedication to do it the hard way.
#2 is the issue that came up in a game run by a friend of mine. Being able to double your money by using half the value in resources to sell for the full price (which is how the game is structured) gives a lot of potential wealth to the players. Now, this isn't necessarily a problem, I don't mind them being able to do this, but it creates problems within the economy. I haven't found a solution I like, though the idea that perhaps you aren't always 100% efficient as based on your roll could play into this, giving some uncertainty into exactly how much it costs to make the item.
And #1 and reason I hate the base 5e system, things just take too long. The base system takes forever. I think... ah heck, I'll grab the notebook and see if I can't find it.
Okay, I'd forgotten this detail. My most recent set of notes had me doing this.
Basic crafting progressed at 10+ ranking die per day, with an additional +10 progress for working in ideal conditions (like a smithy for blacksmithing). The ranking die is based off of the training I mentioned. You start as an Journeyman with 1 proficiency level die. After you have made X gp worth of materials, you become a Master and get a second die, and then Grandmaster, and Expertise gives you an extra die regardless. So a high level grandmaster with expertise would roll 4d12 to add to their roll, a low level journeyman adds only 1d4. This does not roll a d20, because a smith can clearly make a decent sword in normal conditions.
Artisinal work though, was slightly different. Artisanal work had a DC based on this formula (Current Estimated value of the Work / 10 )
So, to make a piece worth 100 gp, with 50 gp of materials would start with a DC 10. If you succeed on the check, you get the base 10 progress + your proficiency times 5. So, succeed on a DC 5 with a +3 prof and you make 25 gold worth of progress towards your goal. If you fail by 5 or more on the check, you subtract the value instead. If you fail five times on a project, ever, then it cannot be worked on any more.
The trick here (and I'd have to dig for if more materials were called for) is that until you failed 5 times, you could keep making the roll. So after reaching a value of 100gp, I could roll again and make it 125, which increases the DC, and I can keep going and polishing the work. You can also start with lower material costs. A mastersmith might be able to make a 100gp artisanal blade with only 5 gp in material cost, with a subsequent rise in the DC of the check.
I'm probably forgetting a lot, I've got an entire notebook and I skimmed to type this much (plus, it is 1:30 AM over here) but I would not be against trying to hammer something out if we could.