Site Statistics
44Site Statistics
aka: Why I'm not fun at parties.
I had a little free time, so I decided to fire up my favorite videogame (Microsoft Excel) and put on my octopus hat and play internet dot-com millionaire for a few hours. Thought my fellow meh'ricans might be mildly interested, or at least leave some slightly amusing snarky replies.
Site Visits by Date
- Daily deals are best on Mondays and Fridays apparently.
- Everybody might complain about Speaker Dock Fridays, but they coincide with periods of high visits. (July 25, Aug 8, Sep 19, etc.)
- Traffic somewhat flat until November. (Was it the email announcing the theme song? Fuku-anticipation?)
- Removed Day 0 and Day 1 (huge outliers).
Site Visits by Day of Week
Notes:
- You guys mostly check Meh during the week.
- You all have lives during the weekend.
- Visits don't necessarily translate into sales, though... or DO they?
Percentage Bit/Bought by Day of Week
Notes:
- Percentage bought is the inverse of "Percent who didn't bite."
- People are more likely to buy in the middle of the week. (Is that where Meh puts the best deals?)
- Nobody buys much on Sundays (is that where Meh hides the crap?)
Other Boring Facts
Products most likely to be purchased in multiples are the cups and glasses with an average purchase size between 2 and 2.58.
The most common events are Speaker Docks (12), followed by cups, glasses, and mugs (11), and then non-dock speakers (10).
Products with the highest percentages who people who bit/bought are Fukubukuros (26.22%), Shirts (15.29%), AA Batteries (4.47%), and Phone/Tablet Accessories (3.48%). Speaker Docks are #14 (1.5%). (This is skewed by sell-outs.)
Quickest Sell-Outs are Fukubukuro 2 (12:20 am), Tenergy Aluminum Power Bank (12:45), The Swilling Knife Set (12:53), Fukubukuro (12:55), and the K'Nex 618 Piece Building Set (1:15). The Solo Active Laptop Backpack was a runner-up at (1:51 am).
I hope you guys found this as mildly interesting as I did. 
- 19 comments, 24 replies
- Comment
Damn- that's some fine work!
awesome work @neuromancer
a couple notes
1. somewhere in the flatlands we included some filtering to try and stop recognizing search/spider bot traffic - it ends up being substantial (thousands of daily visits) now with more real time indexing attempts
2. @shawn updated sellout time stamps to halt conversion tracking at sellout (and post the time sold out) a few weeks ago, so this makes for a huge inflection point on Sellout data.
I love that sharing this data creates interest in Meh as a platform.
@snapster oh @neuromancer - rather coincidentally there's a new load testing routine running that is temporarily screwing up the numbers today - @shawn is tracking down the user agent ID to eliminate the bad data (just for exercizes like this one!)
looks like that data got cleaned up so today's stats are accurate to collect.
@snapster Thanks for the info Snapster. I know that for some of us, seeing how the motor works is as much fun as watching the race.
Nice! I must say, I never got in any fancy graphs when I was tracking numbers over at shirt.woot.
I'd buy more wine glasses. An acquaintance bought a house and they'd be a nice housewarming gift.
@lisaviolet Do you have the data to back that up?
@stupidrobot Yes, I do.
Looking for a job. Got an extra one?
@bhbrigman mediocre.com/jobs
@jsh139 Thanks already been there! :)
Awesome work done here @neuromancer! Thanks for sharing!
Very nice work! It's cool to see an interest in the stats we put out there (from one person, at least).
I hope we can find a way to highlight even more stats without cluttering up the site too much.
@neuromancer fine job! very interesting!
super meh ! today my streak hit 100 !
should I be delighted or dejeted?
maybe just rejected ....
· 118 total meh button clicks
· 100 in a row
· longest streak: 100
@Zebra yes, but how many orders did you order, eh? Isn't that the only important stat for meh? or would that be the dollar amount spent?
@meow57 How many ?? Way the hell too many, according to my wife. Packs of batteries, speakers, shirts, baby arms, cables (I will never use) vacuums, etc, etc, etcrapium, eh!
@Zebra ... Baby arms?
@MsELizardBeth @Zebra baby arm
@jqubed that did not clarify!
@MsELizardBeth baby arm.
@MsELizardBeth Sorry, someone always needs to do that when the baby arm is mentioned. It shows up in the photoshopped product picture from time to time.
@Ignorant well that's terrifying.
Is there any kind of pattern about the prices of items sold? Are there certain days that more expensive items offered?
(I like stats too @JonT. It's what drives me in the Fantasy league you started. Thanks for creating it and introducing me to a new, all-comsuming passion!)
More stats!
More stats!
More stats!
@Teripie I don't see any correlation between prices and day of the week. I didn't put the charts up because they didn't show anything interesting. Perhaps the powers that be spread expensive items out delibrately? or possibly the price vs. day doesn't factor into the decision? If you look at the top five most expensive offerings, about half occur during the week, and half on the weekend.
@Teripie For the record the top five most expensive offerings were: 1. JBL OnBeat Mini AND iPad Air Bundle (440.56), 2. Canon EOS Rebel T5 DSLR (384.25), 3. HP Pavilion Desktop PC (326.80), iRobot Roomba 560 2-Pack (305.97), and Emerson 1080P HDTV 2-Pack (305.87). Note, the lot price was calculated by dividing the days total revenue by the number sold, so includes some shipping and/or tax.
@Teripie Ahhhh but be careful. Correlation does not equal causation… and random shit in can equal random shit out… and type I and II error is a bitch (along with other statistical error like run enough t tests instead of an ANOVA and you will find a false positive just because of chance).
Nice work! Any interesting trivia about the South Korea connection, or does the data parse down to that level?
@editorkid I don't have any inside data. I just went to the sales stats at the bottom of each date and entered them into a spreadsheet manually. I hope the folks running the ship will find ways to make more data available to us. I think understanding how the site really works helps build the community.
This is really cool to see. I do have a bone to pick, though. The 2-for Tuesday Kindle sale sold out twice in one day. The first sellout happened around 1:01am, based on the comments in the thread. They released a second wave of Kindles at noon, and that sellout was 5 minutes later. The sellout time listed skews this product's sellout time, because it lists the second sellout of the day.
@Thumperchick Yeah, I only have access to the stats as they appear on the site. The more samples we get, though, the more accurate the numbers will be in general.
@neuromancer, It seems you also missed the $10 fryer that sold out at 12:15. I wonder what skewed your times?
@Thumperchick Good catch! I entered it as 12:15 PM instead of AM, but we all know whose fault it REALLY is.
@neuromancer Um, who could that be?
@Thumperchick Aw, man, the $10 fryer. I missed that one. starts to cry
I am a sucker for stats too. I think I will translate this and share it with Korean folks.
Excel is my favorite program too! #nerd4life
@neuromancer: My colleagues call me the Excel Czar... I think we must go to the same parties :)
Quickest sellouts is skewed too. Fuku would have sold out in a few minutes had the servers held up.
Is anyone else seeing visits stuck at 24080 for today? It hasn't moved in hours.
@darksaber99999 yep, seeing it over here. we're looking into it.
Nicely done! If you're ever looking to do some analysis involving prices I'm working on a doctorate in mehconomics and also like playing with data.