Poisson Statistics and Ad Performance Performance
One Ad gets a click though rate of 5% and has been clicked on 10 times,
Another has a click through rate of 3% but has been clicked on 100 times.
Poisson statistics could help you determine whether the first ad is performing better or whether it is a fluke of a small number of clicks. Information like this would help advertisers more quickly determine which ads they should invest in.
Re: Poisson Statistics and Ad Performance Performance[ Edited ]
March 2013 - last edited March 2013
AdWords already does that, though it does not explicitly show you that in the inteface.
In the Ad rotation section you can choose if you want to optimize for clicks or for conversions, and the system then compares the ads' performance by looking at clicks or conversions. If it detects that an ad performs significantly better than another, that ad will show more often.
Also, if you create an experiment, and assign one one ad to the control group and one ad to the experiment group, it will show you all metrics, and if there are statistically significant differences between those metrics (CTR, conversion rate, etc.) you will see that in the interface.
Of course, we can also download a report and use a statistical significance calculator and check the confidence interval when comparing the performance of two ads, ad groups, keywords, etc. Of course, manually doing this is not quite fun.
What would be great, indeed, is if AdWords could offer a "comparison" section, where one could add a few elements (ads or keywords, for instance), and compare performance per metric. But I doubt such a feature would be used by too many, and if it would be worth the investment.
Was my response helpful? If yes, please mark it as the ‘Best Answer.’ Learn how here.