Forum Moderators: martinibuster
I purchase PPC traffic for my websites which adsense is one of the revenue streams and I was wondering if there's a way to attribute how much was made from adsense right down to the keyword level of the PPC traffic (eg adwords)
I can do this for actual sales on the webiste, but would like to get an accurate view into exactly where the adsense revenue is coming from versus how much the PPC spend was for a keyword...
anything out there that does all of this or part of it?
thanks!
I would kill (or at least pay) for such a tool!
[webmasterworld.com...]
[webmasterworld.com...]
you can insert the keyword dynamically can't you in adwords (so you don't have to manually add a specific keyword to each different URL)
It's just a matter of figuring out a way how to bring 3 systems together
1. adwords (how much you paid to get the visitor)
2. the adsense tracker (how many visitors vs clicks for which keywords)
3. adsense reports (to find out how much the click on the adsense ad was worth)
I purchase PPC traffic for my websites which adsense is one of the revenue streams and I was wondering if there's a way to attribute how much was made from adsense right down to the keyword level of the PPC traffic (eg adwords)
Yes, kind of. In a general, but very valuable way, and the more formal statistical training you have, the better you can do it. At the most gross level you calculate correlation coefficients using your daily spend results (let's say for one particular PPC traffic vendor - let's say adwords), and your daily adsense income. Vary your spends and you will see the relationship between what you spend, and what you earn.
Presumably you can do this at a much finer level if you want, but the finer you go, the more tedious the process.
I'll actually be running correlation coefficients for the months of Jan, Feb, and March once all the March data is in.
Once you have the data (you download the info in csv format, but you might need to clean it up before analysing it, you can do all kinds of things, like look at the relationships of any and all the variables you might be interested in. For example, is there a relationship between impressions and CPM? Or the number of clicks you bought and income? Or the cost of the clicks you bought and income? Or you could compare the relationships between what you spend and earn for various different PPC sources.
There are other more advanced stats related to multiple regression techniques where you can look at a bunch of variables and how they inter-relate.
The catch is time and effort. Theoretically you could learn a lot about how adsense works, and how to optimize by the numbers. But it's also true that the time it might take you to do it might mean you don't do much else with your time. And the more picky and sophisticated you want to get, the more stats expertise you need.
I have always thought to get some meaningful results with so "fine" statistical analysis you would need samples in the order of #*$!.xxx pageviews per day.
I would be very interested to know the results of your research (I mean, to know if there are results that seem meaningful)!
"Oh, look! there's a Preview button"
I have always thought to get some meaningful results with so "fine" statistical analysis you would need samples in the order of #*$!.#*$! pageviews per day.
Obviously the larger the number of data points, the more confidence you can have in the results, but generally, to get statistically reliable results (depending on how reliable you want them) very much smaller numbers of data points are fine.
So, looking at the correlations between a) what I spend on ppc and b) what I earn will have the same number of data points as there are in a month (e.g. 28, 30, 31 pairs to correlate) if it's done for each month, though you could do them by going across months. Problem is that you introduce other problems if you do it across months, since other changes might have occurred to influence the relationships (e.g. changing your creatives.
While you can do huge amounts of analysis, the ONLY thing I'm interested in is looking at the relationships between what I spend, and what I earn. Theoretically, once I know the correlations, I can predict (all things being equal) ROI for the next month(s).
I'm not interested in statistically analysing data to figure out how adsense works, or going down to the keyword level (that would be a problem for us since your spends are small, and then the data points might be two few.
I wouldn't be surprised if out of the huge population of adsense users, that someone has taken the time to do the proper analyses to try to figure out what google does, but my guess is that if it has been done, that the people who did the analysis wouldn't be sharing the results.
If you calculated the r^2 (coefficient of determination), you could attempt to account for how much of the results are attributable to the spend. 0 <= r^2 <= 1, which will let you assert the proportion of the dependent variable (earnings) attributable to the independent (spend). To belabor the obvious, you could then determine the effect of doubling the spend.
Statistics are fun.
