Forum Moderators: martinibuster
if you have a high CTR, that cut goes up? For instance if the adword buyer pays bids google $5 per click, a low CTR will give you .50 cents and a high CTR will give you $2 per click delivered?
Second, it wouldn't make sense for Google to pay a bigger cut for a higher clickthrough rate unless anticipated conversions for advertisers stayed the same.
but since google doesn't tell publishers what clicks "convert", you have no way of knowing how to improve what you are doing.
all you can do is to give adsense honest relevant traffic, make the ads as noticeable as possible, and sweep up the junk adsense regularly... you are fighting a battle against mfa's, adsense spammers, and ad blindness... your ctr is also affected by how the advertisers word their ads, whether they use text or image ads, etc.
1) Clicks are often not counted immediately. Many of them probably go through a series of tests before their value is calculated. I would guess that there are probably many variations of tests which a click must go through depending on the reference. I am sure Google tracks the referrer.
2) There are more factors here. Many a times a user would click after viewing 4-5 pages. In that case Google probably changes the click pricing(possible)
3) Advertisers who provide conversion data has to be factored into the click value and if adsense uses cookies to track conversions then most probably the value of the click can change over time. So its possible that a click that is marked $0.5 can become $2 in two days and the adjustment is dumped on the current day's earnings.
4) Historical data and smart pricing probably adds more delay in applying the correct pricing to a click. I am not sure but perhaps market valuation is also considered for each keyword.
So overall I think it's probably best to use weekly earnings when you are trying to judge earning differences.
One pattern I have noticed on my site is there seems to be a relationship between sudden dramatic changes in traffic / clicks and epc. If there are slow increases in traffic, clicks and ctr then the relationship between them and epc is pretty stable. So I try to achieve a steady growth rather than try to get sudden spikes where possible.
One recent example would be when I got (temporarily :( )promoted to Google position 1 on my keywords. The increase in traffic and clicks was quite dramatic, though the ctr was rock solid. On the graph for the time, the spike upwards in traffic/clicks was met by a downward spike in epc. Similarly, if I've had a sudden spike in interest for other reasons the same pattern occurs.
I am aware that epc etc. varies for a million and 1 reasons, but the pattern of epc dip when traffic spikes occur is a constant.
if you have a high CTR, that cut goes up
No. CTR has in a very wide range no influence on EPC.
I started AdSense 5th June 2004. I was happy about my high CTR, because all my experiences had been based on less than 0,2% from a banner system, I removed 2000 from my web sites.
I started much optimizing February 2005, and realized, that my CTR had been very low. Now I have a several times higher CTR, but EPC remains in exactly the same range.
Conclusion from the big experiment: CTR has in a wide tested range no influence on EPC.
Btw my earlier CTR was over 10% after I changed the design it dropped below 10%. Maybe 10% is the mark as 2% in adwords?
I think that I side with david_uk's assertion/guesswork/faith that sudden changes in traffic (ie spikes) are discounted (ie regarded as of dubious value) by AdSense (and other ad networks that I use, as it happens).
So when AdBrite accidentally (and repeatedly) delivered me 10,000 new visitors in the 10 minutes after their local midnight (rather more than I had paid for for the whole day in each case) the CPM networks paid me very little extra for all those impressions, and it looks like AdSense's SmartPiercing(TM) decided to discount *all* my traffic even though the new visitors were barely clicking anything at all.
If I were in G's shoes I'd do the same: *sudden* changes are as likely to be problems/bugs/fraud as to be anything good. How many (wannabe) petty crooks (think "how do i make $1000 on my first day in adsens?" threads) reaslise how simple the stats are to weed out these effects?
(Why did I let AB continue dumping the traffic on me? Well, it let me stress-test my system at orders of magnitude higher traffic than I normally get at that time of day... The system sailed through, I'm pleased to say! It was also difficult to turn the traffic off, and cost me a few $100 until I managed too. Bv<)
Rgds
Damon
So from my experience I could state that there are some connection between CTR and earnings.
There's no doubt that a connection exists between CTR and earnings, because (click price) x (number of clicks) = earnings.
The question was whether there was a connection between CTR and the publisher's cut or share of gross revenues.
I meant that after my CTR went down below 10% mark my earnings were cut off dramatically - not propotionally to my CTR.
After that I noticed a sudden CTR drop by 2-3 percent. My earnings were down even by 30 percent. So from my experience I could state that there are some connection between CTR and earnings.
Btw my earlier CTR was over 10% after I changed the design it dropped below 10%. Maybe 10% is the mark as 2% in adwords?
I think there is a problem with your math--you aren't comparing the percentages properly. You didn't give exact numbers, but if you dropped from 11% CTR to 8% CTR that is NOT a 3% drop. It's a 33% drop! A 3% drop from 11% would be to 10.67% (.03 x 11). Do you see what I mean? It's easy to be confused when talking about percentages of percentages.
So that suggests that your 30% drop in earnings was very much in line with your 30% drop in CTR. It actually had no effect on EPC.
Just something that occured to me while I was reading David_uk's post. Could it be that the impressions are added in real time but the earnings are not [because of the reasons I posted earlier] Because if the revenue does not shoot up at the same time as impressions the ecpm will drop. The point will probabl be proved if you see a incremental increase in earnings after the traffic spike.
Did that happen David?
Looking at the graph when sudden spikes occur there is no delay between earnings, increase in clicks and downward spike in epc. As far as I can see, the impressions and earnings are all added that day. Reporting isn't real time in Adsense, but usually by the next day the previous days totals are accurate.
Once the spike is over, epc rises. Except if the spike is a longer term spike, when it appears that smart pricing whacks you with another hit on the way down. It takes a while for things to stabilise again, but they always do. I don't think adsense algo's like sudden changes, and to be honest I can see why it's a feature they might choose build in, even if I might not agree with it.
Btw, I just checked my adsense stats and the CTR change is a bit higher than I stated ~40-50% (not 33%).
I have speculated in the past that it can be worthwhile to remove the AdSense code from pages or channels that are getting very few or no clicks. I've seen my earnings rise when I do it--earnings, not eCPM--because my EPC goes up. I used to attribute this to smart pricing taking CTR into effect. Now I am not so sure. But if you can identify particular pages that aren't working well, and you can't tweak the CTR, then it might be worth trying.
