Forum Moderators: martinibuster
We have watched our traffic levels rise on our network as well and yet still our monthly revenue falls back into +/- 3% of what our usual monthly revenue is.
We basically have stopped trying to increase CTR and traffic as a result of this happening on 5 separate occasions. This is on a large established and respected network of pr 7 and 6 websites a few of which you would know by name producing several million users a month. It is excellent traffic of the best quality that took 5-7 years to attain and not some scraper generated buy-my-viagra-cheap trash.
My resident engineer who is also a math expert told me that statistically it is impossible to have what is happening happen. He explained it to me in this example.
"You have a swimming pool filled with golf balls and each golf ball is worth 3 cents to 10 bucks, now I want you to blindfold yourself and pullout #*$!,xxx number of golfballs this month.
Now wait one year and comeback on the same month and fill that pool up with all new golf balls with totally different values and blindfold yourself and grab x,xxx,xxx golf balls and have them add up to 99.9997% the same amount of money as last years number."
He believes we are being hardcapped to a maximum amount of revenue we can make no matter how much traffic or CTR we increase.
I am pretty savvy and read/post around allot but have not seen to many people bring this up. So my question is has anyone else seen this behavior or is it just us?
Do you think we should just come out and ask G if they have us hardcapped a set dollar amount or not?
Thanks in advance
Even EPC doesn't need to be stable: This month I've seen sharp EPC rises, eCPM rises and overall earnings go up. That was a direct result of changes in the sources of some of my traffic.
OT: If the pool is being filled with balls of random values ranging between $0.03 and $10.00 then no matter how many times you take "x" balls out and refill the pool the balls taken out will tend towards the same total value. The larger the "sample" the more likely it is to add up to your total value figure. I don't know if this analogy he offered you was the best one ;)
I am pretty savvy and read/post around allot but have not seen to many people bring this up. So my question is has anyone else seen this behavior or is it just us?
Stats stuck is more easy to post than stats update normal.
Income down is more easy to post than great increase.
For me is building income by AdSense far more secure compared to internet promotion.
Ad one: CTR 2% and EPC 10 cents. (Overall eCPM =$2.00)
Ad two: CTR 1% and EPC 21 cents. (Overall eCPM =$2.10)
Ad three CTR 1% and EPC 19 cents (Overall eCPM=$1.90)
Ad four CTR 1% and EPC = 10 cents (Overall eCPM=$1.00)
Ad Five CTR 2% and EPC = 5 cents (Overall eCPM=$1.00)
Which ads will google display?
If Google is optimizing for you it should display ad 2. But suppose ad two gets used up or stops performing for some reason. Then google should serve ad 1 and after that ad three. Ad four and five are last choices.
But what do you see?
You see your CTR jump from 1% to 2% and your EPC drop from 21 cents to 10 cents.
After ad one is used up you see CTR drops to 1% but EPC jumps to 19.
Again, it appears that CTR and EPC are being gamed by Google but it is in fact just optimization working for you.
I saw similar things last year in regards to traffic. I have often wondered if I had a 'cap' as well. A floating cap would be one way of slowing down fraud. If I had a cap back then it has been raised. But that is just speculation.
Added: Like you I also saw traffic increases followed by EPC decreases. There was no nice logical explanation for this (assuming I wasn't large enough to be responsible for burning ad budgets), which is whyI wondered about a cap.
As my CTR increased on the Adwords side I was able to lower my min bid often down to 2 and 3 cents.
The efective cap on CPM that you are referring too does apply and that was a good explanation.
My resident engineer who is also a math expert told me that statistically it is impossible to have what is happening happen
He should study economics and the stock market for a while and he'd understand why it CAN happen as simple statistics would indicate I should be making 3X-5X what I am from AdSense but the advertising market doesn't seem to play that way.
He believes we are being hardcapped to a maximum amount of revenue we can make no matter how much traffic or CTR we increase
There is no cap best I can tell (mine goes up and up) as it depends on what the current BIDS are from the advertisers, a supply and demand situation. The more traffic you can supply could exceed the higher paying demand for that traffic. Also take into account how smart pricing impacts your site as your CPC valuation may be dropping due to perceived ROI per click.
Other factors may also come into play but that's a whole different thread.
I've got more than 2 dozen sites running under adsense with one in particular pulling in more than 200k adsense impressions per day.
Just over a month ago we were at CTR of about .8% and had been for over a year. Revenues were at $125-$150 per day. After reading some optimizing tips on how to "blend" ads we made some changes to our main site that boosted our CTR from .8 to about 1.5%. The next day our revenue properly reflected that better CTR and we had revenue of $275+ which would make sense considering the improved CTR. The VERY NEXT DAY, traffic stays the same, CTR still at 1.5% after changes...but our EPC was slammed down enough to bring our revenues right back down to $125-$150 range. Everyday after that, even with the higher CTR's the EPC never was at the same rate.
After a few weeks we did a bit more tweaking with placement and with colors. We pulled a 2% click through and showed good revenue. Again, the very next day, the EPC was down enough to give us pretty much the same revenues we made before the change.
Now here is something that makes things more interesting. We have many sites, that cross over the category (health) of our most popular site. These sites are getting higher EPC for the same ads even though the sites are far less in quality, amount of data, amount of traffic, and even have more affiliate ads than our main site. These sites even have HIGHER CTR than our main site because there is less quality content.
While I can't prove it, it seems that there is definitely something in Google's formula that pulls sites back to their common revenue amount. I suppose one could argue that the bottom simply dropped out of advertisers in that market and that Google was getting less. This is pretty thin argument however because 1. we have other sites getting the higher EPC that have high CTR's 2. the exact same advertisers are on both sites. 3. The site being penalized for EPC is of much higher quality and traffic than those not being penalized even though those sites have higher CTR (low CTR on our good site because people stay on it looking for new items)
Google is playing a middle man between publishers and advertisers without even letting either side know why or how our EPC is being manipulated. While there is little we can do to change this, it doesn't mean we have to be happy about it.
What needs to happen is a union of adsense publishers go together to approach these advertisers directly and cut Google out of the middle. While it doesn't make sense for small sites to do this I think it is worth the time when we are talking 5-7 million impressions a month.
Oh well, I expect the Google fanboys will flame me now for not trusting Google 100%. But I've seen the numbers come and go and IMO there is definitely manipulation taking place beyond site quality and traffic.
My advice to anyone messing with optimization is to take it very, very slow and not shock their formula with a huge surge in revenue. With a few of our other sites we've taken this approach and Google seems to like it. I've grown one site now to almost double CTR and revenue without EPC sliding back by doing it over a two week period. I'm almost certain if I had made the changes all in one night and doubled revenue by the next day, Google would have done some "smart pricing" voodoo to take away the increase. JMHO after seeing the Google game play out.
effectively decreasing the cost of the ads by increasing their CTR?
Does not compute - illogical assumption.
I've downloaded my entire history of AdSense into Excel and generated a column showing the average cost per click and it fluctuates up and down and the fluctuations have no direct correlation to total clicks or page impressions that I can tell at this time.
Some of my highest impression/click days have high CPC averages and some are lower, volume doesnt seem to matter, it's what's available from the advertisers at that time that seems to make all the difference. Bidding goes up and down and some ad campaigns run out of fund or are just closed.
It's all a crap shoot.
I must say that I think the OP is on to something.
Yes, another bunk theory.
Supply and demand is what it's all about, nothing more, nothing less.
His breakdown seems to explain the epc/ctr problem we face rather well!
The only problem I keep coming back to is our traffic is increasing though, normal traffic growth plus increased penetration into new verticals regardless of diminishing returns/increased comp on the web.
That is the part that is still stumping me unless our traffic growth is by some miracle correlating with the eventual downpressure on ad inventories or we are just burning through ad inventories so fast that we end up at the bottom of the barrel .3 .5 cent range which was mentioned and that would not add up to much even with larger numbers like ours.
I'm also an adwords client as well. Adwords are priced per click, not by CTR. In other words it doesn't matter if someone has 1000 visitors and gives one click for 25 cents or has 10 visitors and gives up one click for 25 cents.
As an adwords buyer I tell Google the most I will pay for a click and as far as I know anything to do with CTR is carried on behind the scenes in a black box of some kind.
As an adwords buyer I see NOTHING that puts money back in my pocket when a site like I have as a publisher has it's EPC cut in half overnight. That's part of the frustration here. It's almost like Google has some sort of flag when it sees revenue jump quickly and then they have 'smart pricing' kick in to take a larger cut. From an advertiser perspective we maybe getting something back but it's all in Google's hands what that is if anything.
IMO both sides of Google's model seem to favor Google. At least in terms of what you see on the surface. Perhaps behind the scene advertiers get cheaper clicks and more impressions than what are reported.
I'm personally assuming all this manipulation falls in under the term "smart pricing". My rub is that with 200k impressions and $125-150 in revenue each and everyday they should have already been smart. It seems that only when I started to make more revenue off my traffic that suddenly Google wants to start geting "smart". Not cool, and from my view of the advertiser side it seems to be Google walking away with the extra profits.
One thing for sure is that Google's model makes clicks earlier in the day much more costly. Advertisers set up daily expense budgets and once those are exhausted the EPC's fall. As an advertiser it seems smart to turn off Google in the morning and wait until late in the day to start spending your money.... It goes further.
Well not totally illogical since adwords prices charged to advertisers is based on a minimum value calculation that does include cost of ad and ctr:)
How is it on a Monday following a .8 to 1.5% increase in CTR I get almost double the revenue from $125 to $275 in one day? Never before had we had over $200 in a single day before the CTR change. This primary site and the other sites I own in the same area also have the same EPC for the same set of advertisers and have had for almost a year.
THE VERY NEXT DAY, the EPC plummets 50% on the main site to bring down the revenue to where it was before. Yet the rest of the sites I own with the exact same advertisers keep 100% EPC.
While 200k a day is quite a few impressions are you trying to suggest that this ONE site's increase in CTR somehow now sucks enough ad inventory out of Google to drop 50% EPC across the entire market? Yet somehow the other sites in the exact same industry STILL HAVE THE HIGH EPC and it's that way day after day, night and day, rain or shine!
This is far more than global supply and demands across Google network. I could believe your theory of supply and demand only if I didn't have three other websites getting 50% more EPC for the exact same advertisers day after day.
IMO this is "smart pricing" that somehow discounts sites they think they can stick with lower EPC. It seems IMO to kick in when Google spots an immediate and drastic increase in revenue.
You had an excellent post earlier, but that is not what happened in my case.
You said...
"But what do you see?
You see your CTR jump from 1% to 2% and your EPC drop from 21 cents to 10 cents. "
In my case the EPC stayed the same and the revenue doubled after the change. It wasn't until the day AFTER the huge increase in revenue that the FOLLOWING DAY the EPC was cut in half and remains cut in half. Other sites in the same niche with less traffic did not see an EPC drop.
While I certainly am familiar with the ups and downs and fluctuations of EPC there is definitely something drastic that Google has done to plummet the EPC after we increased CTR AND made strong revenue off that increase.
THE VERY NEXT DAY, the EPC plummets 50% on the main site to bring down the revenue to where it was before. Yet the rest of the sites I own with the exact same advertisers keep 100% EPC.
It's hard to speculate on such swings without seeing the all the details of your site.
I've seen HUGE swings on my site in both direction and I've usually been able to determine the cause. Sometimes it's just one keyword I find in one channel generating a bunch of money that I theorize some advertiser accidentally did with too broad of an ad campaign and it went away quickly when he saw the bleed on his credit card.
Other days it's VERY minor fluctuations in CTR as 1% +/- in CTR can swing my earnings as much as $100/day and so can changes in the advertisers bids.
So far I've always found a reasonably rational explanation for these types of trends and a good use of channels will often times help pinpoint specific spikes and dips.
I'm convinced that my EPC plunge was directly related to fact we doubled revenue literally overnight. Google had no problem delivering the ads and paying for them at full 100% EPC for that ONE day after the change. But I think when Google got the numbers at the end of the day the site was quickly targeted for "smart pricing" or whatever they wish to call it.
In my experiments with other sites in rasing revenue through CTR SLOWLY, I haven't had the EPC plunging problem. In other words EPC stays up with small 10%-20% daily increase in CTR by "blending" a few pages at a time. But I have a feeling that a 100% increase in revenue overnight could possibly trigger some sort of negative re-action on Google's part.
It is sort of like the care some are taking in adding pages and submitting them to Google. If you add 50k pages in one day through your database you risk the possibility Google may think something funny is going on even if it legitimate content. Add 50 pages a day and you won't have a problem.
JMO
I was wondering if this pricing stuff was site specific or it blanketed multiple sites? I manage several sites using channels and it seems that my primary site eCPM keeps dropping (from lower CPC apparently) but other sites that get a lot less traffic seem to get much higher CPC for almost identicle ads.
it seems that the channels that get very little traffic get much higher CPC - im wondering if I should expect the CPC to drop if traffic builds?
In my own case I saw from January 2005 to April 2005 earnings rise from hundreds to several thousands of Dollars only to be spanked down very hard and overall earnings are only just now returning to those levels even though, like everyone else, genuine traffic etc has increased.
I started experimenting and decided upon a new formula for my 100+ web sites. I now construct region specific trade widget sites and product specific trade widget sites instead of promoting the all encompassing mega trade widget/product site.
By this I mean widget.cn, widget.in, widget.co.uk, widget.de etc. The widget.com's still exist however I try to use those sites to lead the user to the satellite sites.
This has had a profound effect on earnings which are now rising a steady rate rather than exponentially plus the CTRs, eCPM's have to be seen to be believed.
The mega sites are still earning good money and rising gently however I am convinced my new strategy is far more effective plus I feel I have one huge bonus...IF the SERPS were ever to destroy one of my mega sites there's a pretty good chance all the satellite sites will survive and they all rank pretty well too.
Don't forget, it's just my experience, I have no idea whether I am correct!
For me the EPC plunge happened with just the site I had optimized and doubled the CTR overnight. The other sites did not show a drop in EPC at all. It's also important to note that we did receive full EPC and full revenue for a full day before it was whacked. That is what makes me think it has less to do with inventory of the advertisers and more to do with Google "smart pricing", whatever that really is.
I think you will see the more clicks the lower the EPC will be even if all is going well and you have not been smart priced. This was explained well by others in this thread.
Thats the best suggestion I've heard in some time. If Google are capable of it, I wouldn't put it past them.
Another way to predict EPC might be to keep an eye on competitors. When YPN beta came out my EPC went up for a couple of weeks. When everyone pointed out it wasn't that great, EPC went down again.
Chitika are a popular replacement for Adsense this past week. Ooh guess what? EPCs up by a third.
Not terribly scientific, but I find it amusing. Might be worth an experiment or two. Perhaps whoever is most cheesed off might like to start a thread praising their amazing earnings from Yahoo this month?
On a practical note yesterdays eCPM was high and we had the highest earnings ever, Result today the eCPM is slashed by 1/4.
I also plotted eCPM vs date and the eCPM is still on a slope of ~0.5 rising with decreasing variation which I would attribute to some form of genetic algorithm trying to find a higher return. Fitting on that was a spline. A Friedmann shows something equivalent of the famous rollercoaster.
So empirical says something is going on, stats [albeit basic at this stage] say everything is great..
In reality you will have multiple variables going into this, maybe you sould do a model and some test scenarios. Most of what is posted here, also by me, is _often_ emotional guess work. If you have a maths guy on hand why not let him set up some kind of more scientific tests?
We basically have stopped trying to increase CTR and traffic as a result of this happening on 5 separate occasions. This is on a large established and respected network of pr 7 and 6 websites a few of which you would know by name producing several million users a month. It is excellent traffic of the best quality that took 5-7 years to attain and not some scraper generated buy-my-viagra-cheap trash.
If your sites have a "relatively" limited set of keywords and your user base is that big surely there must be some maximum that you will move towards. If you are already at that maximum a result as you describe could also happen and would explain the small error in this. If there is no more money in these keywords that would be it I assume. Maybe you just maxed out that Google algorithm in your niche, even if it is a big niche. Surely there must be a maximum a publisher can achieve also depending on bandwidth, backbone, Google ad delivery speed, Adwords of a certain kind fed into the system. On a relatively large niche with a huge sample size I guess the maximum would also be quite stable as in having low variation. Hard to know without more facts.
This would be the only other logical conclusion beyond being hardcapped for total possible revenue yield. As I understand the adwords systems more as was pointed out earlier in this post the more I get the sinking feeling that we are basically ripping though all available inventory for our verticals.
I am getting the basic picture of the following
1. we increase ctr% and the only inventory left that we are increasing into is low paying .3-5 cent adverts
2. the more we increase traffic the only inventory left that we are increasing into is low paying .3-5 cent adverts
The only piece left for me to figure out is why when we penetrate into a new vertical with in theory a totally different set of keywords do we not see a increase in our total account rev from that new traffic. Our whole dollar amount ends up coming back to +/- 5% of what it was.
The only piece left for me to figure out is why when we penetrate into a new vertical with in theory a totally different set of keywords do we not see a
increase in our total account rev from that new traffic. Our whole dollar amount ends up coming back to +/- 5% of what it was.
Ends up means over what time frame? If it's the same than that's indeed suspicious. :\
Maybe we should try Sherlock Holmes.. [exclude all ... etc] or the famous
Occam's razor:
Given two equally predictive theories, choose the simpler.
;)
My income seems to keep going up but those two things stay the same, neck and neck, not identical but very very close. When earnings are going up they both rise a tiny percent....It's puzzling to me but I am happy with it.
Ad one: CTR 2% and EPC 10 cents. (Overall eCPM =$2.00)
Ad two: CTR 1% and EPC 21 cents. (Overall eCPM =$2.10)
Ad three CTR 1% and EPC 19 cents (Overall eCPM=$1.90)
Ad four CTR 1% and EPC = 10 cents (Overall eCPM=$1.00)
Ad Five CTR 2% and EPC = 5 cents (Overall eCPM=$1.00)Which ads will google display?
If Google is optimizing for you it should display ad 2. But suppose ad two gets used up or stops performing for some reason. Then google should serve ad 1 and after that ad three. Ad four and five are last choices.But what do you see?
You see your CTR jump from 1% to 2% and your EPC drop from 21 cents to 10 cents. ...
creepychris gets the idea but that is really not what does happen in the Google's auction system. The writer has been mistaken between the Max. CPC and actual CPC. Max CPC is used for auction (what ad to be shown first) while actual cost is used to set eCPM. Max CPC can be the same or 10 times the actual CPC. By the way, actual cost is also based on the performance of ad below.
For example:
Ad three CTR 1% and EPC 19 cents (or max CPC)-> Bid CPM=$1.90 -> position 3-> eCPM=$1.01 (one cent above 4th position ad)->EPC= $1.01/[1%*1000)=10 cents
Thus, Max CPC=19 cents but paid 10 cents
I have found on repeated occasions with Google AdSense that:
1) Google HATES change. This includes large swings in traffic in the upwards direction. If I get an abnormally high traffic day - My EPC plummets... I don't think it would matter if I got 10x more clicks than on a low traffic day - payout will still be nearly the same.
2) After sustaining this increase in traffic over the course of a couple weeks, income begins to steadily rise - but it will never grow linear to your traffic for the vast majority of sites... if not all sites.
I am one who does believe that there are filters to hold you in place to an extent. When traffic and clicks are up 4 fold, and EPC is reduced by approx 75% as a result - my red flag goes up, and so does my middle finger.
