Forum Moderators: martinibuster
just my 2 cents for this great comunity
What I really want to see is a breakdown of impressions, clicks and revenue by page. Can that really be so difficult to do? Do you suppose Google is afraid of abuse of some kind (although I can't imagine how the data could even be abused)? I would think it would be a big win for them as well as publishers and advertisers. If I knew which pages were generating the most revenue, I would certainly create more of those kinds of pages!
Using SSI allows thousands of pages to be altered at once. Allocating a channel for every page [and I don't think you can anyway] is nearly mission impossible.
If it is not possible to get data by URL, how is google monitoring specific pages to see if things are invalid (high ctr, whatever)?
Channels are usefull to some degree, but not getting data in real time makes the whole idea garbage to me.
Channels were a very very very VERY poor attempt at better tracking. I hope that is not the best they can do..............
Well that explains what my problem is. I don't think G knows that in business time is money. If it takes more than 12 hours these days to do any kind of a report, it's usefullness is very limited.
URL data is 100% useless for understanding which ad format or position produces better results.
URL's would be a shortcut solution for some but for most web marketers they would be inferior to channels.
What other way would you have to test such things on the exact same audience within the exact same environment at the very time of using these varying appearances and still read the user behavior?
Channels are excellent and as a first AS improvement (of many more to come) an excellent first step and choice. I didn't expect anything less from Google though.
What I need, but am not getting, is stats that show me which pages generate the most clicks over time. That would allow me to build more of those kinds of pages, which would improve my income as well as Google's. I've heard of 3rd-party tools that do this. Does anyone know if there are freeware versions?
What I need, but am not getting, is stats that show me which pages generate the most clicks over time.
This can be done in aggregate. With over 100,000 impressions it sounds like you have a great many pages. With a great many pages it sounds like an inefficient way of judging the efficacy of ads.
Another approach is to judge blocks of pages according to what kind of page it is, to assign it to a group and judge the epc fingerprint of different ad modules to the different classes of pages, and thus tease out the appropriate ad module for the appropriate page. To drill down to a specific page may not be the right approach if you have many pages.
Real time stats. would be pointless for reaching any conclusion. Just because computers are quick doesn't mean decisions have to be and it certainly doesn't follow that quick stats. are better because they're quick. Its not a computer game!
The number of variables involved in the daily revenue and the level of daily variation just make it impossible to judge the statistical significance of any daily or even weekly variation.
Except of course if you make a change and your channel revenue multiplies by say 5.
It makes it very, very slow to make changes and see the results of said changes. It has nothing to do with the speed of computers, but more of an issue of feedback.
When one tunes a radio broadcast amplifier, one adjusts a variable capacitor till the power gets to the desired level, if the current is too much, then one must adjust a coil till the current is in spec. If this causes the power to drop, then one must go back and re-adjust the cap. If I tried to do this over a 24-36 hour period, well, to say it is not very efficient would be an understatement. It is the same principal as with the reporting.
I have about 500 pages. If most of my clicks are coming from a handful of pages, that is important data that I can use to increase the value of my site. With more relevant content, everyone wins.
Yes, I could use channels to do this by chopping pages up into sections, and measuring each section separately, then zooming in when needed. But in my case, where the majority of the site is built around a single template, that's just way too much work.
The radio analogy is interesting, but that is trying to tune some variables to a stable setpoint.
Here, we have the problem of tuning variables to maximise another variable, where some of the variables are under your control and some are not (and not even known for sure).
The only way of knowing whether what is done has worked is by using statistical significance testing. Doesn't mean hiring a mathematician, but does mean the periods of tuning are quite long.
So if you've got a set of data collected which shows the distribution of revenues over a period where you made no changes, you can then make a change and then monitor the significance of any variations in revenue over the same period.
Its like trying to spot a real signal over noise.
If you are in the situation of having a massive number of impressions, you make a change and revenue doubles for the next few days then the significance is pretty obvious. But for a lot of the time, the tuning period will need to be longer.
The reason I posted is that its too easy to sit in front of the PC and make changes thinking that they are having some effect. You make what you think should be a beneficial change and for the next few days the revenue goes up by 30%, so cause and effect appear to be known. But then it heads down again. I've done it!
Reminds me of the old 55 MPH speedlimit. When they first made the change, there were fewer accidents. Many years later, back to 65/70 MPH, and the same thing happened -- fewer accidents. Cause and effect are not always obvious.
So does the average visitor not really know or care anything about what the Adsense ads. are? After all, there's no reason they should know that the ads. vary and are targeted, unless they study them every day.
