Reposted from a month ago on the ARS message board by one of our users Argy: Lenghtly post but a lot of REAL statistical analysis involved.
Okay, since this comes up on a daily basis, I'm going to try to illustrate some effects of random distribution. Say you get 1500 hits a week, and average 1:300. So you should get 5 signups a week, right? Maybe 3 on an absolutely terrible week, 7 on a great week. In actuality, you're quite likely to see greater aberations once every couple months. Here are a three computer simulations of 26 week periods:
Week 1: 5
Week 2: 2
Week 3: 3
Week 4: 8
Week 5: 5
Week 6: 7
Week 7: 6
Week 8: 0
Week 9: 4
Week 10: 4
Week 11: 4
Week 12: 7
Week 13: 7
Week 14: 5
Week 15: 7
Week 16: 3
Week 17: 8
Week 18: 6
Week 19: 5
Week 20: 6
Week 21: 2
Week 22: 11
Week 23: 5
Week 24: 5
Week 25: 4
Week 26: 4
Total: 133 (1:293)
You see a week 8 had 0 signups. That's when people squawk "what's wrong with ARS, you suck all the sudden." Then week 22 has 11 signups. That's when people dance the ARS jig. Note that this has nothing to do with ARS servers - these are totally random distributions. Here's another run:
Week 1: 8
Week 2: 2
Week 3: 7
Week 4: 7
Week 5: 5
Week 6: 5
Week 7: 4
Week 8: 5
Week 9: 4
Week 10: 9
Week 11: 5
Week 12: 1
Week 13: 1
Week 14: 1
Week 15: 7
Week 16: 8
Week 17: 6
Week 18: 8
Week 19: 6
Week 20: 6
Week 21: 5
Week 22: 5
Week 23: 9
Week 24: 4
Week 25: 9
Week 26: 6
Total: 143 (1:273)
Whoa, three weeks in a row with one signup, ARS has surely gone to shit! On the other hand, the overall conversion ratio is quite a bit better than the 1:300 probability, so this is actually a lucky streak when taken as a whole. Suppose this webmaster always griped on the message board when you got only one signup in a week, and suppose ARS has 100 webmasters with the same kind of traffic. Every week, four webmasters are going to gripe that this is their worst week ever, that they usually get 1:300, and suddenly they've crapped out at 1:1500. Think about those numbers - ARS has a lot more than 100 webmasters, and a lot of people who have lower volumes than these.
Now for something different. Suppose you're still getting 1:300 conversions, but can send 30,000 hit a week, so figure on roughtly 100 signups a week.
Week 1: 106
Week 2: 92
Week 3: 109
Week 4: 105
Week 5: 112
Week 6: 103
Week 7: 92
Week 8: 104
Week 9: 95
Week 10: 98
Week 11: 100
Week 12: 99
Week 13: 105
Week 14: 94
Week 15: 87
Week 16: 96
Week 17: 105
Week 18: 98
Week 19: 107
Week 20: 91
Week 21: 101
Week 22: 92
Week 23: 99
Week 24: 90
Week 25: 114
Week 26: 108
Total: 2602 (1:300)
Wonder of wonders, the numbers look a lot more stable. (It actually is unusual that it hit exactly a 1:300 average, although it's usually going to be 1:295-1:305 with this volume). Again, think about how many people who usually get 100 signups a week come on here to bitch about "ARS has gone to shit this week." It happens on occasion, especially when there's a real problem, but by and large the big sellers aren't griping because their volume gets the numbers closer to their averages.
It's something to think about next time you think ARS is letting you down.
------------------
ARS Up to $44 per Signup. 13 New Sites including Voyeur Dorm