So, the 3 month journey was trying to identify the best dungeons to farm for transmog while removing as many preconceptions in the testing process as possible. We got 30 people together on stream to do 20 runs of every dungeon Vanilla through WoD, used DBRegionHistorical to calculate value, and took an average run. This basic setup of "Run the dungeon. Add it up. Average it out." has some problems though.
- We only did 20 runs. This was just a limitation of the channel size. It already took us 2 weeks and 200 Once I'm giving Pewdiepie a run for his money, maybe we'll redo the test for 100 runs of everything.
- It took us two weeks. The goal was to get a snapshot of prices, and since the testing took a little while, prices may have varied over the course of the test. Probably not enough to worry about though, but worth mentioning.
- DBRegionHistorical is a mean. Means can be highly skewed by the extreme prices, and we saw some price shenanigans with the WoD dungeons (but that was probably a no-competition 110% DBMarket feedback loop). This is probably fine though. Since we're comparing dungeons, we don't actually care about the price, just the relative difference between two farms. For Vanilla at least, it's reliable. I would have preferred to use a Median but TSM doesn't support that baseline anymore. The TUJ addon has global medians, but that would have required volunteers to download more addons, and I wanted to get as many participants to help as possible.
- DBRegionHistorical is a regional price. Some people had the nerve to play in EU! Prices are a little different between the two realms, and may have affected rankings slightly, but prices are hardly ever 2x as much.
- We took a mean lap. A crappy farm where you can get absurdly lucky will have the same average lap as a really good farm that's super consistent. Luckily since we recorded each lap's value, we could calculate a median lap.
- Lap time matters. Durnhold keep has the same lap value as ZF but takes 4x as long. So I ran and timed a full clear of each dungeon to calculate a median gold per minute.
- We only did full clears and didn't test raids or heroics. Repeatability and consistency of testing was important so the lap speeds of a normal farm might change is you avoid certain bosses. i.e. Halls of Stone full clear is 11 minutes, but a 0/4 60% trash clear takes 60 seconds. I used my farming knowledge to retest any spots afterwards that had obvious skips that would make a big difference.
- Both mean and median matter. Median gold per minute is good, but if two dungeons have the same median "goodness" but one can drop super awesome expensive stuff, it's a better farm. So I ranked everything my mean gold per minute, then by median gold per minute, added up the ranks (to avoid weighting by huge differences in gold value) and reranked that total.
After all these calculations, I think we have the best possible way to calculate what people think of as "a good farm." Now we can rank them.The Results
There were some fun results. Just in the vanilla stuff, the popular farms like Zul'Farrak, Uldaman, Scarlet Monestary were all in the bottom half. I put the top ten vanilla farms in the video. There were some wild result though, like, Violet Hold (WotLK) and Oculus were both as good as Uldaman on a per minute basis. Only one of the top ten were actually vanilla.
I made 505k profit after 3 months of posting. Tried to do every day but fell off about halfway through. Got in between 50 and 70 posts per item. 80% of my sales were vanilla, 50k profit on BC, and while I never hit 0 gold on my Wrath through WoD posting toons, I did lose money there. I blame my posting operation though. It was pretty greedy with a minimum price of "min(.75 x DBMarket, .65 x DBRegionMarketAvg)". Still, this helped support our theory that the prices for expansion mog is artificially inflated. It's not that it doesn't sell - we showed that you can sell 1/3rd of your inventory in a week if you post Iron Docks gear at between 100g and 2000g - but it does show that it doesn't sell at the current prices of 130k gold for a lame chestpiece.
My theory is that because of the 7.3.5 and 8.0 changes, we have less competition for this gear. When there's only one poster they set the DBMarket price. So if they're posting at 110% DBMarket, every post is a little bit higher than the last. If enough people do this because of TSM default operations or popular imports, then eventually the people posting at regional prices are also affected. Make your own operations, guys. Don't import. Learn how the addon works and use it to mirror your selling mindset.
I've heard people recommend to vendor anything with a sold per day of 0.01. 60% of my items sold were 0.01 items. Imo, they're worth it. 15% of my sales were rings, necklaces and trinkets. Blues sold slightly better and for a much higher % of the DBRegionHistorical. Farming specifically for blues is probably a bad idea, but it might be worth it to be more greedy with them in your posting operations.
The Posting Fees
I've always said posting fees are negligible. Posting fees are 30% of the Vendor Price per 24 hours posted, and for vanilla gear that comes out to like, 30 silver for armor that sells for 2k. For vanilla gear, I made 450k, posted 37,000 auctions in 90 days, and spent 45k on fees. Items took an average 60 posts per sell, which is a little skewed because not everything sold. My average gold earned per post was 12g. But this ignores farming time.Posting Cost - the most important result of the TTT
To calculate how much posting's worth, I first calculated how much time each post took. 2,500 items took 1 hour 15 minutes, so each post took 1.8 seconds. With 60 posts before an item sells that's an extra 3 minutes of posting per item.
So, if you have a farm that gives you 40 items worth 100k in one hour, you then have to spend an additional 2 hours of posting time before everything sells, and that brings the final value of the farm down to 33k/hr. I think this is the most important thing we learned from the tests and is the missing piece to why the expansion stuff ranked so highly. It's not that it won't sell, and it's not that it won't sell at those prices. But posting time starts to matter.
Gold farmers haven't ever had a way to calculate sales time, and now you do! If Iron Docks or BRC stuff gave us 40 items worth 500k, then it's the best by how we calculate vanilla dungeons. But if it costs an average of 8g/post and it takes 2000 posts before it sells at the inflated DBRegionHistorical prices (2000 posts is a made up number), then that's 16,000g per item in fees and 60 minutes of posting time per item. If you value your time how most people do, at what you get from 1 hr mining+herbalism (currently between 10k and 20k) then that's a 26k-36k posting cost per item. 40 items x 30k cost each is 1.2 MILLION in the combination of real and opportunity costs. So the farm is worth 500k in revenue but 1.2m in expenses, so it's worth negative 700k. If you post the items more aggressively though, in the 100g-2000g range, it's a 30k per hour profit farm, but then it doesn't make the top ten.
So, there's a lot more I cover in the vid, but I learned a lot here. I'm excited to have numbers but I hate misleading people and the sales results imply that our initial ranking calculation method was wrong for expansion transmog. In case this was the result, I made sure to include farming strategy in each dungeon vid, so even if the ranking is off I still think the videos are important tools for learning. New farmers need to learn ability management, how to look for density, speed, taunt rotations, and valuation. The testing really did a good job at clearly showing that the value of a transmog farm, in general, is the density (enemies killed per minute) times the average item value. If you want a better farm, either find something with more density, higher mean item value, higher median item value, or any combination of the three. Final note, unfortunately, the TSM ledger is currently swollowing up 2/3rds of the sales data so some of my numbers are estimates, but there's a lot to be learned regardless.