Forum Moderators: martinibuster
Anyone else seeing similar?
My ctr is down a bit but ecpm has improved a lot. Although it will probably get knocked down by the end of the day.
But if google wants to click dump, then Ill be happy with the updation ;)
Yesterday was another record day for us, though. Given this morning's slowness, perhaps we got an advance click dump from Google?
1 out of the 5 are related to my content.
Yep. Many of us are experiencing this problem and are pretty ticked off about it. My month has been ruined by poor ad targeting. CTR is way off, which of course means earnings are down to pitiful levels as well. It started on Oct. 1 for me, but others have mentioned it began in Sept. I sure wish YPN was out of BETA and firmly established so I could vote with my feet and leave AdSense in the dust. If enough of us did it, maybe they'd figure out a way to fix this obvious problem. They don't seem to care at the moment (they won't respond to my emails about it) - probably because they know there aren't a lot of comparable alternatives just yet. I bet they get a LOT more responsive to us publishers once YPN and maybe MSN have services that truly compete.
Swebbie, my memory may be playing tricks, but I thought you had already signed up to YPN, and had had some success. What is it that being out of BETA will give you - more advertisers?
Also, again my recollection may be poor, but I recall you saying that you banned some advertisers. That may be contributing to the poor ad relevancy you are now seeing. Have you tried removing all the bans to see if things improve?
Regarding bans, I do have a few in place with AdSense and with YPN, but they're really only for sites with content I don't like or are directly competing. I'd prefer a bit less income to showing those ads to my visitors. On a related note, YPN is doing a very poor job of actually implementing ad blocking. I've talked to them about it, and they admit to problems there. That's why they have BETA periods I guess.
What a pity since Monday to Wednesday were cracking days and on target for one of my best weeks since May...so I'll have a meal and vino collapso on G tonight and hope that spurs them into action:-)
I first noticed the change on Thursday, and it's continuing today.
Ad targeting seems to be on point. Wonder what's up.
don't chat up any foreign birds - according to the news they might have avian flu...
The one I was with definitely had wobbly legs by the end of the evening...I thought I was on an ocean-going liner in heavy seas the way she kept sliding up and down the bar with absolutely no assistance other than liquid refreshment!
Stats still look pants...page impressions are way off my logs, CTR down 26% yesterday and 14% today and now the EPC has also taken a nose dive, 15.93% down on average.
Ad targeting seems to be on point. Wonder what's up.
Yep, I can't complain about that either, unless I have a mega click dump at day break I won't even make a weekend earning...most strange considering this is an extremely busy time of the year for our trade.
Ever get the feeling that something new is about to be announced/launched/etc? I say that because every time I see this kind of statistical anomally G invariably does exactly that.
Swings, roundabouts and rollercoasters...
I've been signing the praises of Fastclick and I think I am beginning to understand why my AdSense revenue is actually increasing as I'm adding as many as three (a top 728 or 468 banner, a skyscraper and a small rectangle.
While Fastclick is providing steady CPM revenue (and an occasional boost via some CPC clicks), their ads are often garish and bold and always graphic, while the understated design of Google's ads may be more amenable to clicking, and of course, they are targeted, while the Fastclick ads are not. The Fastclick ads get about a .5% clickthough, the Google ads upwards of 2.5%.
So, for now, all is pretty rosy in my little corner of the world. That is sure to change, as I know from eperience.
The EPC remained in the pan and the eCPM the lowest in 6 weeks.
Today? Boinggggg...everything back to normal with already 27.7% of yesterday's earnings with 7.9% of the visitors and the CTR and EPC bang on where it usually is at this time of day!
What did they do?
There must be a very strange feature in the program that allows such aberration.
djulien - Have you bounced back?
"Statistical anomaly"? What you describe doesn't sound to me like a statistical anomaly, it sounds like statistical normality. Using our stats as a basis for analysis (but that's making big assumptions, I know) our standard deviation of CTR is 10.6%. What that means in English is:
Whilst your sites may vary by that much I can assure you that mine do not and never have done in the past couple of years.
Whenever something like this has happened there have been significant changes by Google.
your CTR being down 26% is a big drop, but you'd expect it to happen, on average over the long term, once every three weeks
Why do you say that? Is that what happens to you? Certainly it is a very unusual occurrence for my sites which then raises flags when the other statsistics are analysed.
being down 14% is also a reasonably sized drop, but you'd expect it to happen almost twice a week
If that occurrence were on a regular basis then I would agree with you however, once again, for my sites it is a most unusual situation.
My sites are statstically almost "boring" in their lack of fluctuations except when G are changing something or there is a problem at their end.
Work is hectic enough without G's algo playing mind games with me:-))
It's the laws of statistics. Standard deviation is a way of measuring and anticipating variations in data.
Raw figures (eg: one day being down by 25%, say) are in themselves meaningless. In my own case, for example, Friday's earnings were 45% down by comparison with Thursday. Am I panicking? No. On the contrary, Friday's figures are reassuring because, when put in the context of the right statistics, they are quite "normal".
To say of any figures that they are not normal, without doing proper stats, is a psychological interpretation, irrespective of the figure involved. This doesn't mean that there isn't some substance underneath it, but to work out whether there is something to be concerned about, or you are unnecessarily worrying about anomalies that are actually part of the normal ebbs and flows of statistical variation, you have to understand something about statistics and do some analysis.
Standard deviation is useful because it provides an objective, quantifiable measure of what your normal range of variation is, and how often you are likely to get figures outside that normal range.
[robertniles.com...]
I have no association with the site. I hope this URL doesn't get cut because all of the "authoritative" sites out there are written in gobbledegook. This one seems much easier to understand and has a nice, clear picture.
I don't wish to get into an argument about this however I had better clarify that my entire working life, 37 years now, has revolved around statistics and I am a bit of a geek when it comes down to it!
I fully appreciate what you are saying and the Robert Niles page is a good starter for those that have never studied it.
Whenever I see a huge deviation from "my" expected norms then I delve further and ask if anyone else has seen a similar occurrence just to ascertain whether it's my stats which look weird or whether there are several with a similar problem.
Running 100+ sites means it takes me a bit longer to nail down the problem even with the assistance of Channels, however I have found one of the culprits which was one specific site's Adlinks.
That site's Adlinks EPC yesterday was down by 50%. I have check every day since September 1st and not one day was less than USD 0.20 however yesterday it was USD 0.115.
Why? I have no idea however today it is USD 0.244. Did I have that many clicks yesterday on that one specific site and everyone of them hit the "cheap" button?
Statistically, most unlikely BUT possible...
Interestingly, bearing in mind all the sites are niche widget sites carrying the same themed ads, none of the other sites Adlinks EPC was down, they were all in my accepted "normal" range.
The last time I saw this level of EPC it actually spread across all my sites and I had to do some drastic culling and reorganisation of the sites, however I am putting this one down to Google doing something in the background, hence the big difference between their page impressions and mine, and which will become evident very soon.
Yup. I'm rebounding with a vengeance. My CTR and eCPM are back to the same respectable levels I enjoyed prior to Thursday. Things are better than ever.
OptiRex: "Well, I didn't get my click dump for breakfast ... G's overall impressions were down about 15% compared to mine."
Firstly, click dump for breakfast? Sounds heinous no matter how you say it. :P
This has to be an age old issue, but why would the reporting be skewed negatively for us? I'd hate to think that this was deliberate.
My eCPM levels for the past couple of days had been at lows that I hadn't seen for roughly four weeks.
Any possible explanations if my traffic levels and page views are otherwise phenomenal and ad targeting is spot on?
Now how about plugging away on this data with some Poisson theories... ;)
Thanks for the additional background. It's useful to know, because when you read these inert pages on the screen, the impression one gains can be different from the underlying reality, because of lack of context (it's often impossible to tell whether someone really understands their data or not, or their level of understanding of stats). And I agree with your rationale of asking about odd numbers to ferret out any problems that occur (one of the two advantages I cited yesterday of watching short term stats).
With your stats experience, though, I'm sure you agree that, when presented on their own, percentages are nevertheless meaningless, and can even be misleading. The number of standard deviations from the mean would convey a much better sense of the scale of the problem.
I agree that the longer the date range interval, the more likely you are to pick up discrepancies.
Throw in a wildcard variable like G updates, and there's simply no way to feasibly trend on issues.
Over a long period you may be able to calculate probability of a G update occurring in a given month.
Has anyone actually attempted this?
[edited by: djulien at 4:18 pm (utc) on Oct. 15, 2005]
>It's useful to know, because when you read these inert pages on the screen, the impression one gains can be different from the underlying reality, because of lack of context
Yes. Its always worth remembering, that. Before one goes off on one, so to speak. :)
I'm sorry, but I don't understand the point you are making, here. Whether Google make sporadic changes has no bearing on the relevance of standard deviation as a yardstick for measuring the degree of change. In fact, there are often lots of factors that constantly change underlying a dataset. In Adsense terms, you also have: no of advertisers, market rates, new publishers, etc. that change, in addition to Google algorithms.
To be useful, standard deviation does not rely on data being constant. It is simply a unit of measure, like an increment on a ruler or tape measure, to provide a sensible way of assessing your data, in relation to your own experience/data. Eg: to a novice long-jumper, 5 metres is a long distance. To an experienced athlete, 5 metres is a short distance. Over time, however, the novice's definition of short and long may change, depending on recent performances.
In a similar way, although one's Adsense data may change over time, standard deviation simply helps to giving meaning to the raw, meaningless numbers by putting them in the context of (recent) performance. Even if standard deviation were rendered irrelevant because of some unusual characteristic of the data, how much more irrelevant does that make individual percentages?!
I need an updation!
Ann
This has to be an age old issue, but why would the reporting be skewed negatively for us? I'd hate to think that this was deliberate.
No I don't believe it is and I feel there is a simple answer to it which I have asked several times however no one seems to be able to come up with the answer(s).
Where are the Google Adsense data servers held and how many do they have? Why do I ask this?
I have two Adsense accounts and they update statistics and react completely differently. Usually Adsense 1 account sees occurrences several days before my Adsense 2 and "most" others see changes in Google's algos and its modifications.
Maybe our accounts are on this main "test" server and possibly data is deleted, unavoidably delayed or since they are possibly using an intense server load, it simply cannot be written to the program.
Several thousand page impressions, along with their associated clicks, just do not disappear over a couple of days when "our" stats have been in agreement for the past couple of months. I just feel they never even get written to the database, it's a darned niggling "fault" and, as you say:
Yup. I'm rebounding with a vengeance. My CTR and eCPM are back to the same respectable levels I enjoyed prior to Thursday.
In my industry there is no such thing as short term advertising or a significant number of new or retiring advertisers, my advertisers are constantly there every day of the week, month in month out.
It really would be nice to know the answer since here I am on a Saturday evening having already easily surpassed Friday's earnings and everything looking like "regular" business.
Does anyone have any insider knowledge about this?
