4 ms·

http://en.wikipedia.org/wiki/Simpson's_paradox http://en.wikipedia.org/wiki/Simpson's_paradox

by cgs1019 13y ago

IBM 13y ago
I'm not too smart, you're going to have to explain the relevance of that to my comment.
- raldi 13y ago
  Consider these hypothetical numbers: 2012: 1 million desktop clicks at $1 each + 100,000 mobile clicks at 10 cents each = $1,010,000 total cash, or 0.918 dollars per click, overall. 2013: 1.5 million desktop clicks at $1.10 each + 2 million mobile click at 20 cents each = $2,050,000 total cash, or 0.586 dollars per click. Overall clicks went up. Mobile clicks went up. Desktop clicks went up. Overall cash went up. Mobile cost per click went up. Desktop cost per click went up. Everything's great! Except if you're just looking at overall cost per click, it looks terrible.
ariwilson 13y ago
The best example of this I can think of here is desktop and mobile clicks. Mobile clicks have historically had vastly lower CPCs than desktop and are increasing in quantity faster than desktop clicks, but I can easily believe that both mobile and desktop CPCs are individually going up.
ChuckMcM 13y ago
Except you'd really want to figure out the confounding influence. The GP is correct, that Google's CPC has gone down for 10 of the last 12 quarters. It is reported in the quarterly earnings as a total across all clicks (which may be why you think the paradox applies) but by comparing revenue from properties and sites one can deduce that in this case it actually reflects that Google's ads are becoming less and less valuable. That has been more than made up for though by Google increasing their "inventory" which they can do up to AltaVista levels if they choose, but that gets another paradox going which is Ad fatigue. My read of their results over the last 3 years makes a pretty compelling case that the Ad business peaked 4 years ago and has been fading ever since.

4 ms·

http://en.wikipedia.org/wiki/Simpson's_paradox http://en.wikipedia.org/wiki/Simpson's_paradox

by cgs1019 13y ago

http://en.wikipedia.org/wiki/Simpson's_paradox http://en.wikipedia.org/wiki/Simpson's_paradox

IBM 13y ago
I'm not too smart, you're going to have to explain the relevance of that to my comment.
- raldi 13y ago
  Consider these hypothetical numbers: 2012: 1 million desktop clicks at $1 each + 100,000 mobile clicks at 10 cents each = $1,010,000 total cash, or 0.918 dollars per click, overall. 2013: 1.5 million desktop clicks at $1.10 each + 2 million mobile click at 20 cents each = $2,050,000 total cash, or 0.586 dollars per click. Overall clicks went up. Mobile clicks went up. Desktop clicks went up. Overall cash went up. Mobile cost per click went up. Desktop cost per click went up. Everything's great! Except if you're just looking at overall cost per click, it looks terrible.
ariwilson 13y ago
The best example of this I can think of here is desktop and mobile clicks. Mobile clicks have historically had vastly lower CPCs than desktop and are increasing in quantity faster than desktop clicks, but I can easily believe that both mobile and desktop CPCs are individually going up.
ChuckMcM 13y ago
Except you'd really want to figure out the confounding influence. The GP is correct, that Google's CPC has gone down for 10 of the last 12 quarters. It is reported in the quarterly earnings as a total across all clicks (which may be why you think the paradox applies) but by comparing revenue from properties and sites one can deduce that in this case it actually reflects that Google's ads are becoming less and less valuable. That has been more than made up for though by Google increasing their "inventory" which they can do up to AltaVista levels if they choose, but that gets another paradox going which is Ad fatigue. My read of their results over the last 3 years makes a pretty compelling case that the Ad business peaked 4 years ago and has been fading ever since.