Forum Moderators: martinibuster
September saw an 11% reduction from August's EPC
October saw a 5% reduction from Sept.'s EPC
November saw a 23% reduction from Oct's EPC
December (1st two days) is seeing a 20% reduction from Nov's EPC
By my calculations there has been a 45% drop in EPC from August.
My site is very broad-based (directory site), so these results are likely not from a single industry's advertisers reducing their budgets, etc.
I would have expected a jump in EPC as we approach the holidays, as advertisers bid for more traffic.
When reading claims of increased EPC on this board, I get concerned -- though I'm sure results are mixed. Even with more than double the amount of traffic I had in Aug/Sept/Oct I'm still seeing less payout.
Any ideas, or is this just a sign of the times?
I do get about 15-25% PSA/Alt. ads which does drive down earnings, but would not affect EPC.
While that is also a small sample size, one month, according to your other months, that may be your high.
[edit]but the sample is too small to have any confidence in that evaluation[/edit]
>>I do get about 15-25% PSA/Alt. ads which does drive down earnings, but would not affect EPC.<<
I have seen that if the page names change, then you get PSA ads until mediabot sniffs it. There may be other reasons to get PSAs, new pages, etc. But 25% would be a killer.
How do you measure # of PSA's?
I posted a thread about that sometime ago: [webmasterworld.com...] There was quie a lot of response to that thread. Jenstar suggested it could have been caused by Broadmatching. But it seems that a lot of publishers who haven't been aggresively giving Adsense more pages - and better locations - have been seeing decreased EPC. It could well be a drop in the % Google is sharing with publishers. There will now no doubt be some posts suggesting there is no proof that there is a change in pay-out rates :-)
is this just a sign of the times?
Hummm, that’s interesting. I have taken all my competitor ads out. Why? Well there are good sound business reasons to. There are reasons why you do not see Wal-Mart ads in Kmart and Burger King ads in McDonalds. You do not want to allow your competitors to be able to do branding at your expense. You do not want you potential customers to even think about the competitors. Politicians do it all t he time by saying things like, “my opponent’, or “the other candidate”, etc.
>>There will now no doubt be some posts suggesting there is no proof that there is a change in pay-out rates :-) <<
Well that goes without saying.
[edit]I assumed that G is aware of these basic business principals, but, perhaps not.[/edit]
I posted a reply to a similar question a while back. There are many guesses as to why this is happening and without direct answers from Google they are hard to confirm. This is one proven theory.
As an AdWords user I made the following changes to my account.
Here is an example.
I had a daily budget of $100 and was paying $1 per click. My daily budget was exhausted every day. I decided that since there were so many AdSense publishers that I would decrease my maximum ppc from $1 to 50 cents. I found that I doubled my Adwords traffic.
Why would I continue to pay $1 per click when I can get twice as many clicks by cutting my maximum ppc in half.
I am not saying this is happening with every advertiser in every industry but I am sure that there are quite a few advertisers doing it.
I am sure that any publishers site that got any clicks from me a month or two ago are getting half the amount today.
Just for the record I also use AdSense so advertisers doing this are also affecting my income.
novice
November EPC decreased 65.2% compared to August.
EPM has also dropped for me, though not in the same pattern.
September EPM decreased 35.7% from August
October EPM increased 5.7% from September
November EPM decreased 56% from October
November EPM showed a 70.1% decrease in EPM compared to August.
The change in EPC and EPM was dramatic and sudden as of November 1 and very evident in my daily earnings.
CTR is stable. Alternate ads remain stable at about 5% of daily impressions (5% of page views, not Adsense impressions).
September impressions increased 6.9% from August
October impressions increased 8.3% from September
November impressions increased 31.1% from October
November impressions increased 51.8% compared to August, over essentially the same number of pages (I only added 5 pages during that time, all in October, those page receive minimal traffic).
The total earnings pattern is similar to the EPM pattern.
September earnings decreased 31.3% from August
October earnings increased 14.5%
November earnings decreased 42.3% from October.
November earnings decreased 54.6% compared to August.
Hummm, that’s interesting. I have taken all my competitor ads out. Why? Well there are good sound business reasons to
Quite simply, this has been a good month and we've been turning away orders because we are so busy. Allowing competitors ads on the site means we actually get money for turning away orders :-)
For the really profitable own-brand products we do customers will look at our competitors' products and WILL come back. Everytime. ;-)
As more sites put Adsense on more pages, the supply of ad space is growing. Combine that with the fact that advertisers come and go, depending on their daily budgets, and the net effect is to give lower bidders more exposure.
Ergo: average Adsense EPC drifts downward, as a natural result of how the system works.
And Google is more interested in how full their coffers are than yours. They're not a charity organization as far as I know, an neither am I.
I've seen a significant reduction of AdSense earnings since joining in the summer. At first, I tried to keep pace by adding AdSense to more pages and sites. Finally, I decided it wasn't worth the effort to keep chasing the initial revenue levels until I see the earnings swing the other way. But the changes come as no surprise for the reasons stated above.
Lucky you. And eveyone has their own way of doing business, If I had too many orders though, I would figure out a way to fill them before I turned them down.
>>Ergo: average Adsense EPC drifts downward, as a natural result of how the system works<<
I can see that happening, but I question the rate that it is happening since the last update.
I couldn’t believe that I had **** number of visitors according to adsense.
AdSense does not provide visitor stats. They provide the number of times their Javascript code is activated on your site. The AdSense FAQ explains why this number does not correlate with your server logs.
But it would be within a certain margin of error. It's just a question of what that margin is, and what is the average delta and the extreme delta. Using these stats over a certain period of time could indicate how accurate their stats are.
I can see how a page with heavy asp, JAVAScript, etc. could slow down their js a lot, and thus make them make such a statement, but, if you have a very plain and simple page, there just isn't a bunch to slow down the execution of their js.
If your visitors do not have Javascript enabled or they are using a browser that does not support Javascript, nothing that you or Google can do is going to change the fact that those page views will not be counted in your AdSense Impressions.
Based on what I have seen, and that is all I am talking about, there are very few users that have js turned off. That is just on what I have seen due to pop-up pages on my site that use js. These pages contain an explanation of terms and the number of times they are activated. If their UA doesn’t support JS then they have also changed the name of their UA to IE, etc. for I’m not seeing any UA that do not support js. While there could be one person doing this, I doubt that 100’s are. Heck 92.55% are known js UA, IE, etc. 3.61% Unknown/Other and 2.4% Cache/Proxy server (MSProxy) 1.3% Cache/Proxy server (Unknown/Other) and I stopped counting all the .01%, but they looked to be mostly js browsers.
richmondsteve
>>statistical jargon are irrelevant<<
It’s not just jargon. Statistics are always relevant and the ‘jargon’ is required to know what area of statistics one is talking about. Statistics is the bases of pretty much everything in the business world. And since I have been using that tool for 18 years, it is the one I know best, so it is the one I use the most.
There is no doubt that there is lots of room for error, to use some more jargon, which is why I keep saying that the sample size is too small. There is a point where the sample size becomes big enough that the error is no longer relevant. Just because you may know some ‘jargon’, doesn’t make you an expert to tell me that it is ‘irrelevant’. How long have you been writing statistical reports for Fortune 500 companies?
Though not as statistically adept as many here, I'm able to plug numbers into a spreadsheet and the one I study most closely is the daily EPC. At this rate, Google will catch up with the affiliate programs by mid-2004, meaning it won't be worth bothering with.
My background is in engineering and if you read many of my posts here you'll find that not only do I understand statitistics, I do statistical analyses frequently and bring up statistics regularly in this forum. For example, these 3 recent threads I posted to.
[webmasterworld.com...]
[webmasterworld.com...]
[webmasterworld.com...]
I don't know what your sample size was, but mine was 6 figure impressions across 2 sites and multiple days. I adjusted the timestamps from my logs to be in the same timezone as the Google reports (PST), then I substracted alternate ads (I don't use PSAs) by totalling those from my stats and impressions without JS enabled (again, tracked on my end) and the variance between my numbers and Google's was under 0.1% (yes, 1/1000) with all of the data combined, no more than 0.2% when looking at each site individually on a single day (lowest sample size for a site for one day was 5,000 impressions).
Again, YMMV, and you don't know me from a hole in the wall, but I'm confident in the accuracy of the Google reports.
Nah, what gave that idea? ;-))
Day h*ll, it started a week ago.
I understand what you are saying. But I don't know how accurate you data collection is. Heck looking at six digit numbers alone is 100% inspection and, according to research at Northwestern University, has only a 60%, and I have measured worse case 40%, effeteness. As I said I’ve ben doing this a very long time and I know how one collects data is the 1st step in getting accurate numbers in the end. I’m NOT saying that G stats are AFU’ed all the time, but it is without a doubt that their system will lose some numbers from time to time. Heck, I know that from just running banners from my site. You will also find periods were they are worse than others. And while you were looking at 1 month, week, or whatever it was, just 6 digit numbers cannot imply accuracy. Example, you looked at Nov. and there were 100,000 hits, (six digits), but in Sept., they were having server or SQL problems. Hence, there is not enough data to have any confidence. Right?
It is highly evident that the EPC side of the equation has been declining. It was discussed in great detail here,
[webmasterworld.com...]
and although we hoped at that time it was a result of broad matching introduction, seems to have continued through November from the posts in this thread.
Although this may not be affecting all at this time, I think arguing over statistical evaluation, versus reality of the above equation is academic at best.
Believe the obvious, people are seeing lower returns over the last couple of months. Since the program start epc's have decreased in many markets, and apparently the trend continues. Preople are not quoting minor changes on day to day fluctuations but % drops over months.
The second side of the equation of clicks is also obvious, you have less or more. Factors such as ad blidness, number of PSA's being served etc can affect this. However the posts indicate this is not the major impact.
So lets all agree that many are seeing an erosion of EPC....it is a reality to many....and hopefully it is not a trend that continues much farther.
In case you're curious, here's how I did my calculations.
1. Recorded impressions for Sites A and B for a given day from Google's reports (both sites on separate accounts).
2. During the test period I added a call to a small JS file that's purpose was just to track whether JS was enabled in the user's browser for each page view. I made the assumption that if this file wasn't called then neither was the AdSense JS and the requestor either was a human with JS disabled (or a browser with no JS support) or a machine (bot, etc.). I removed the JS tracking code after the analysis so unfortunately I can't test any recent data.
3. The AdSense code was setup with the alternate ad paramater to load an HTML file on my server whenever a PSA would otherwise have been shown.
Since my servers are GMT-5 (well, I think they were GMT-4 during the test due to daylight savings time) and Google's AdSense reports were GMT-8 (GMT-7 at the time) I copied my log files and wrote a script to update the timestamps to change the date/time to use the same timezone as the AdSense reports.
I calculated the # of impressions I would expect to load AdSense by taking total impressions from my log files and subtracting the numbers from steps 2 and 3 above.
Then I compared those numbers with those in the AdSense reports. Mine were consistently less than 0.2% higher on a daily basis per site (2 sites on 2 separate servers).
jim_w, if you have any criticism or concerns with the process let me know because I'll probably analyze again in the future and if I need to abandon it or improve it to make it more accurate I will. Keep in mind that my goal was just to see if anything looked fishy (I was skeptical of the impressions reported at the time), not to prove it was valid at a 99% confidence level. :-)
Has anyone else done a similiar analysis? If so, what were your findings?
Got some data here that indicated this is not happening. Stats indicative that 96.458 % of pages served also were counted by google. This is over 5 months of data. We did a check and feel confident to say that more than 3+% of visits were by known spiders. Has to start you wondering about how accurate googles claims in this area are?
This is readily apparent when both the googlebot and slurp are crawling the site heavily, and the impressions to pages served are within .5%.
