Forum Moderators: martinibuster
For me November was a extremely good month. I was getting EPC in the $2.5 - $3 range, clicks were high, earnings were consistently at record breaking levels - life was good.
1st December rolls round and suddenly EPC drops down to the $1 - $1.5 range, yet traffic and clicks remain about the same (roughly).
What’s going on?!
I can take a look at the website with the preview tool, but the biggest country of visitors, isn't mine, so the tool only gives you an approximation of the ads being displayed. I can ban any known offenders (like ring tone ads etc) but would this just be filling up the filters unnecessarily?
Can you good people offer any suggestions?
Thanks
This could be the explanation as to these types of threads, and the fact that people are always either below or above "average" in significant numbers - all for the same time periods, roughly. I go through periods all the time where the eCPM/earnings are fairly high, and then for no apparent reasons (no site/traffic/ad changes or changes in any other metrics) BANG, for a week or two, the eCPM/earnings are cut in half. Then, mysteriously things go back to "normal" again. For a while, that is.
The only thing that remains a constant when this happens is that when the earnings drop, the quality of ads seems to as well.
[edited by: Scurramunga at 9:21 am (utc) on Jan. 4, 2007]
That isn't going to happen (how many affected publishers will even be aware of your proposed strike?)
Your drop in eCPM could be caused by any number of things that are more likely than your being singled out by Google for economic punishment.For example (in no particular order):
- In the computer sector, buyers--and many advertisers--may be hanging onto their money until Windows Vista is out and hardware innovations such as multicore CPUs and hybrid hard drives are more widely available. (I'm holding off on my own desktop and laptop purchases for exactly those reasons.)
- In the electronics and photo sectors, there way be a lull as buyers (and advertisers) wait for product announcements from the CES and PMA shows.
- In the skiing and snowmobiling hardware and vacation sectors, ad demand may be down because of a lack of snow in many regions.
- As I suggested in an earlier post, landing-page quality scores on the AdWords side may have forced some advertisers out of business or pushed them toward other networks. (I've noticed a decline in ads from affiliate hotel-booking sites, which could be seasonal but may also be a consequence of changes at AdWords.)
- Finally, we're only halfway into the second business day of the new year, and some advertisers haven't finalized their 2007 ad budgets while others may simply not have gotten around to placing this year's ads. (Today's EPC and eCPM are looking a lot stronger than the first two days of the month, although it's really too early to draw conclusions. It'll be interesting to see how the entire month of January looks compared to January, 2006.)
EFV, these are all valid-looking and reasonable possible causes, but they can be excluded because of the -50% figure.
What -50% figure? Any number of publishers have reported different figures and, in some cases, improvements. Of course, a conspiracy theorist can suggest that there's a rotating pool of -50% publishers, but that doesn't make it so.
In any case, there isn't much point in debating this endlessly; whether Google is pocketing a bigger share of the overall take will be easy enough to see when the 4Q 2006 earnings report comes out.
I think I've figured out the recent decrease in CPM/EPC.
Google is applying the quality score to content sites. One of the main factors is if you have adsense on the page. If adsense is on the page then the page gets a low quality score, followed by a smart pricing hit.
So, the only way to not get smart priced is to remove adsense :) We should all switch to ypn or something for a week.
Back to my rum & tonic.
Google is applying the quality score to content sites. One of the main factors is if you have adsense on the page. If adsense is on the page then the page gets a low quality score, followed by a smart pricing hit.So, the only way to not get smart priced is to remove adsense :)
I know you're joking about the "If AdSense is on the page then the page gets a low quality score" part, but there may be an element of truth in that. The presence of AdSense in itself can't be correlated with poor quality (after all, sites like THE NEW YORK TIMES and THE WASHINGTON POST use AdSense), but certain AdSense practices almost certainly can. It all comes down to profiling--which may be reason enough to avoid AdSense overoptimization or trying to suck the last drop of milk from the cow.
efv, where is your "overoptimization" term defined by google? give us a link that proves the existance of such a theory.
google goes to great lengths to educate publishers on how to optimize the placement and blending of adsense blocks... and they put up over 10 ads per page on some of the search results.
perhaps you are confusing it with borderline fraudulent page design, where some publishers do things like make adlinks look like page navigation.
from what i have experienced, one problem is with putting adsense on every single page of a site, without bothering to use channels to monitor the results... it could be driving down the epc for the entire account.
>> where is your "overoptimization" term defined by google
it isn't, then we would know exactly where to draw the line and that would be showing their hand.
It is a valid concern but no one can really say at exactly what point you cross the line
the term "overoptimization" is indeed tossed around in discussions about seo, but that is not relevant to adsense.
Also, let's not forget that Google needs inventory for low-bid ads, not just for high-bid ads. If you've got three ad units on a page with 50 words of text, that may or not work to your advantage, but it certainly works to Google's advantage, because the second and third ad units allow the display of more (including lower-paying) ads.
It all comes down to profiling--which may be reason enough to avoid AdSense overoptimization or trying to suck the last drop of milk from the cow.
I think it's very conceivable that they could and would do this. I wouldn't be surprised to find out that they did.
no, because a lot of adwords advertisers do not give conversion data to google.
which means that since google can't always use conversion data, there could be many types of profiling going on... but that is not relevant to "adsense overoptimization".
i guess that efv wants us to believe that google itself is engaging in "adsense overoptimization", as he calls it:
"If you want the biggest revenue impact for the smallest effort, we recommend optimizing your color palettes. Choosing the right palettes can mean the difference between ads your users will notice -- and click -- and ads they'll skip right over."
[google.com...]
What -50% figure?
agreed but I don't know what other term would apply. If we talk about 'optimizing ad placement' or 'optimizing a site for AdSense' then the term 'overoptimization' would also apply. there are tons of terms I believe are misused but if I don't use them, or can't rebrand it, then no one will know what I am talking about.
>> there could be many types of profiling going on
I'm sure of it, one of which could be what I am calling 'overoptimization'
Suffice it to say that publishers who understand the relationship between price and value are likely to achieve higher earnings per click than those who don't.
unfortunately that rather nebulous way of thinking leads to fud-mongering, as we've seen here.
for example, some people didn't have any moral issues with the pic-per-ad page design, and in fact adsense reps personally approved hundreds of pages that were designed exactly like that... then google changed their mind, and we saw some holier-than-thou attitudes out here on the subject.
there is no such thing as "adsense overoptimization", but there is definitely borderline fraudulent(or worse) page design.
>> attempt at defining
I use the example of the word spam ;)
>> unfortunately that rather nebulous way of thinking leads to fud-mongering
fair enough
my december 2006 epc was nearly double what it was in december of 2005... it was my best ever adsense month, with adsense getting a lot less traffic than it got a year ago.
my epc might have gone up because i spent all year removing adsense entirely from pages that it didn't perform on.
and btw, i "optimized" the h*ll out of every new page that adsense was on, in part using the advice that google gives us.
so suffice it to say that publishers who don't understand what page/site/account optimization really is are likely to see decreased epc.
I'm not sure there is such a thing either. However, when the great Adwords LPQ thing started.. I decided to at least guess at what might be over optimization of adsense since i think my site has good content otherwise.
I used to have ads right in the visual hot spot, splitting the text into two sections. I stopped that practice completely by widening the text and having a smaller ad square float on the right side so there was no break in reading. On pages less than two screens long I removed all adsense from center text and placed a block in the right nav section. I also removed adsense from key pages such as the home page.
The result was apparently no improvement in 'quality' but my CTR cut in 1/2. And now my CPM is <25% of what it used to be.
I've read elsewhere on the net that your site sux if you have ANY ads in the body of the page.
Why i still see ads for the best x sites on a few pages annoys the google out of me.
so suffice it to say that publishers who don't understand what page/site/account optimization really is are likely to see decreased epc.
Or, try this more simple approach. Reduce the number of impressions from 30-50% and your CPM and EPC will rise significantly. Doesn't matter which 50% you get rid of provided, of course that you don't run your CTR into the ground.
This alone would account for Dan's increases.
(and I bet that's much more accurate than Dan's interpretation).
dead wrong... rbacal, there is no statistical relationship beween epc and impressions.
google has never made that claim, either, so i don't know how you came up with an idea like that... but i guess that anything can happen when people don't understand the significance of epc, and they never use channels to monitor it.
sailor, what happened to your epc when you made those changes?
My epc has plumeted by 40% over the last 10 days.
That was preceded by a plumet of 60% this summer/fall compared to year-ago figures.
So, decreasing CTR significantly has had no positive affect.
Actually, if I had to guess, I'd say that removing ads on pages with low CTR used to make for better performance overall.
I think (hope) the recent decline is year-end budgets and slow startup of new budgets. I also think there is generally less advertisers in my niche.
I think too that folks are quickly moving to Y-overture with their new interface (which is super by the way). I'm adding a campaign a day now.
If epc/cpm don't improve next week i'll likely turn on ypn ads (again).
dead wrong... rbacal, there is no statistical relationship beween epc and impressions.google has never made that claim, either, so i don't know how you came up with an idea like that... but i guess that anything can happen when people don't understand the significance of epc, and they never use channels to monitor it.
Actually, the idea comes from looking at data, and anecdotal reports stretching back YEARS from webmasters who have observed that there are diminishing returns from traffic increases. If your traffic/adsense impressions double, for example, you almost certainly will NOT double your income. As impressions rise, there's substantial evidence to indicate that cpm/epc drops. The opposite is also true.
Why? Who knows. But the phenomenon is quite real.
And, it explains, in a manner applying occams razor, what you've experienced. And for that matter what we've experienced this last month, and over years. Impressions drop. Income doesn't drop proportionately. Over the last two weeks or so, our impressions/traffic have dropped between 40-60%, but our CPM/EPC has increased so much that actual revenue is just about the same as with the higher traffic.
In any event, the test is simple. Run a statistical analysis correlating impressions with cpm/epc. You'll get an inverse relationship and negative correlation coefficient. If I ever reinstall my stats software, I'll probably run the analysis on our own data.
no, rbacal, i just checked the exact numbers, and my total monthly impressions were down by 25%, not 30-50%... so your unsubstantiated theories are not relevant to my sites.
>>>most of them correspond to a couple of pages, sometimes rarely visited. Strangely, those pages generate the higher EPC on my site, always<<<
i've seen that also, but rbacal wasn't referring to rarely visited pages... he's making grandiose claims about account-wide traffic vs. earnings patterns, and it's a fond sentiment that has been expressed out here by other publishers.
did you notice how rbacal failed to quantify even the most general stats? for instance, at what traffic level will everyone's earnings start to trail off? a thousand imps? a million imps? does it matter what sector you are in? is it affected by the phase of the moon, lol?
I would expect this to happen in a very small niche as a result of the better paying advertising income being consumed. However, I see this happening consistently across a wide area with no specific topic receiving a particularly large number of visitors.
I've noticed increases in EPC whenever a problem or action has caused my traffic to fall significantly. This begs the question as to whether I could optimise my return by limiting my traffic in some way.
[edited by: Pengi at 9:14 am (utc) on Jan. 5, 2007]
BTW, I'm not sure where the "overoptimization" nonsense came from. This thread has always been sheer shooting in the dark, but this is too much already.
BTW, I'm not sure where the "overoptimization" nonsense came from.
Maybe from the fact that smart pricing is based on Google's estimate of whether a page is likely to convert, and pages that are optimized for clickthroughs at the expense of conversions (e.g., by placing three blended AdSense units above the fold, as scrapers often do) are likely to result in clicks that aren't motivated by an interest in what the advertisers are selling.
The goal of smart pricing is to bring the net price of a click into line with the value of that click to the advertiser. There's a fine line between optimizing for visibility (e.g., placing a leaderboard at the top of content or a rectangle within an article) and over-optimizing for "clueless clicks" (such as making ads appear to be content and burying the real content below the fold). Publishers who cross that line have only themselves to blame when smart pricing, quality scoring, or some other adjustment mechanism catches up with them.
