Forum Moderators: martinibuster
I Track everything I can to try and maximize profitability. The only thing I can't track is CPC... This bothers me ALOT! Right now, the only acurate sample we have, is a per channel per day CPC. NOT ENOUGH!
Ok, I'll try and keep this simple so let's start with one page. On any one page, PV after PV we have a myriad of ads being displayed by G. When we're lucky, someone clicks on an ad. At this point I know what Page, IP, Destination URL, and KW used if it was paid traffic.... Still with me?
What if at this point (the click) I assign 1 POINT to the Destination URL. I do this all day for all the clicks for that page and at the end of the day, I divide the earnings according to the ratio of points.
Given a large enough sample, this could average out to a pretty close GUESTIMATE as to what a particular ad earns. Then divide by number of clicks to get eCPC for an AD. hmmmmmmmm...
Ok now what to do with this new info.? Well for one, you could weed out low earning ads with the filter and in theory this could potentially take care of MFA's at the same time.
Any Ideas? Feel free to poke holes in my theory. (I assure you I can take it) Also, if this sparks any ideas, Feel free to expand on it.
If an ad pays well may depend on many factors. You might have one ad paying a wide variety of epc depending on time of day, location of click in comparison to location of advertiser. You also have to take into account ad budgets running low and Google stretching the budget by lowering the click price, if the month has an R in it, the wind direction and loads of other factors none of know about.
I'd concentrate on increasing serps ranking and pulling in more organic traffic and booting off the MFA's instead :)
You also have to take into account location of click in comparison to location of advertiser, ad budgets running low and Google stretching the budget by lowering the click price, if the month has an R in it, the wind direction and loads of other factors none of know about.
lol yes I had thought about the R in the month... Along with all the other unknown variables. However, wouldn't that just require a larger sample to average them out? Take smartpricing for example: Goggle does not track conversions on all Advertisers, however, they take an educated guess as to your conversion rate. For all we know we convert very well with a few advertisers admist the many we don't.
the problem is that you haven't associated a specific ad with a specific click... all you are working with is a specific ad block that typically rotates advertisers throughout the day.
channels give you enough info to judge the performance of a specific ad block, but that's the limit of the info that we'll ever get from google.
1) Create for this page a channel of its own
2)at this point (the click) create a new channel
etc...
at the end of the day you will know exactely how much is the CPC for each of the clicked ads
the problem is that you haven't associated a specific ad with a specific click
With my click tracking scripts, I Do track what particular Ad is clicked. So at the end of the day, I know how many times that Ad has been clicked.
2)at this point (the click) create a new channel
Wow... my script ain't that good. :p
Take your 200 channels away from whatever worthless pursuit they are currently engaged in. In the page of interest (a single ad unit, for simplicity), display your ads using channel #1 until you get a click. At that point, immediately start using channel #2. And so on, until you get to 199. Then just start using #200 for everything, while you wait for the final Google stats to arrive in your account.
Combining with your auxiliary data, you should be able to get fairly exact payouts for individual ads. Some data cleanup may be required, as more than one separate click might arrive on the same channel. Can probably just discard that channel from the analysis; the fact that 199 >> # of available ads helps. Cross-verify your findings by seeing whether every click on ad X paid the same (well, sorta, since advertisers can vary their bid by time of day...).
Won't work real well if you're averaging dozens of clicks per minute on the target page, but if that were the case, you probably wouldn't be posting here anyway :-).
Wow... my script ain't that good. :p
well it is not hard to do
create x channels and save their numbers
then tell the script that after each click to alternate the channel number in the specific page
Ads don't pay the same per click. What we get per click can vary widely for the same ad given the factors of time of click, location of click and many other factors only Google know about. The solutions mentioned seem to regard the epc of an ads as a constant - it isn't, so it may lead to blocking advertisers that generally pay well based on a few low value clicks, or blocking MFA's based on a couple of OK(ish) clicks.
does it record the exact text of the ad at the time it was clicked? how do you know that the advertiser didn't rotate his ads? some ads are more effective than others.
does it tell you how many ads were in the ad block at the time that the ad was clicked? even if you had that data, you can't control the ad block configuration... so you'd be filtering advertisers based on incomplete data.
>>>So at the end of the day, I know how many times that Ad has been clicked.<<<
as david pointed out, the most you'll ever have is an average epc(not cpc), in part because google does not update it's reporting in a timely manner.
But none of these solutions takes into account that one specific ad may generate $2.00 for one click, but be worth 3c to the next click depending on where it's from and how likely G's guesswork thinks it is to convert.
Actually, by using one channel per click, you should be able to see any such variability directly. I would guess that for many ads on many sites, there will be little or no intra-day variation in payout.
Google can't figure out what it will do any longer - hence all the updates. If they don't know what it does any longer, it seems it's not a lot of point us expending effort to analyse it in any depth.
Not really accurate. Google will not know how a change will affect any given site, page, etc, or other small subset. It CAN know how it will affect the across the system performance, via modelling.
