Update: Guess which site on Thursday came back with its earnings to within $ 0.04 of average?

Dunno about Feb. 3rd, but the 10th was the day of a huge snowstorm in the eastern US, which shut down several major cities and knocked down power lines in many places. IF your visitors are mostly US that could be a factor.

Were impressions down too?

Were impressions down too?

Nope, all metrics normal except for AdSense.

I keep mentioning to you guys that the data from a single day is meaningless. You will never find an explanation that satisfies you. You'll save yourself time and energy by at minimum looking at the week as your unit of time, or better yet, at the month.

There isn't even a guarantee that the numbers for Feb. 3, don't contain data from the 2nd, or that data from the 34d isn't displayed for the 4th or 5th.

You will never find an explanation that satisfies you.

Yep, we know that since G would never admit to a data loss however doesn't it even intrigue you when the days either side have completely normal metrics and that it has affected only one site of several plus you have to admit a 60% earnings drop is pretty dramatic.

I'm not wasting any time on it, I just wondered whether anyone else had seen anything similar?

Yeah, this is a bit different from the usual "stats stuck" issue. One advertiser pulled all ads, then put them back?

Yep, we know that since G would never admit to a data loss however doesn't it even intrigue you when the days either side have completely normal metrics and that it has affected only one site of several plus you have to admit a 60% earnings drop is pretty dramatic.

A lot of things intrigue me, but I prefer to look at things I might actually be able to have answers for than on things I can't possibly get answers for.

I suspect many veterans of adsense have learned that it's pointless trying to figure "it" out.

It is what it is and what ever it is is in a black box somewhere over "there".

Also statistically speaking the odds are you WILL encounter a rare day which is WAY out of whack. And as people have said over and over to you and others, to determine what is a chance occurence, versus something that needs looking at more carefully, you need to do proper data analysis, using standard deviations, etc. Even averages can't tell you.

Also statistically speaking the odds are you WILL encounter a rare day which is WAY out of whack.

And just to prove YOUR point Saturday's earnings were my best since 2006 at 55+% over average and mainly because of one specific URL channel averaging 125% over its normal EPC and that channel was already my best EPC anyway.

Sunday returned to normal...ah:-(

Guess what?

It's Wednesday and guess which site is having yet another disastrous Wednesday so far?

This is getting a bit too coincidental to say the very least.

Well, your clicks are down every Wednesday on that site. Is your traffic down or about the same? Maybe you have a weird advertiser that either advertises only on Wednesday (and nobody clicks on it) or who pulls his advertising on Wednesday (so nobody *can* click on it) ? Does the traffic from that site come from the same places as your other sites? It does seem a little odd. Do all your ads have separate channels on them, so you can figure out if it's ALL ads that are getting fewer clicks, or just certain pages or ad blocks?

Traffic is looking absolutely normal, this one site so far today has seen 25% of an average day's clicks, 50% of average CTR, 36% of average eCPM and 17% of average earnings.

Advertisers appear to be the same, certainly my biggest paying 728 x 90 graphic banner advertiser is there.

Traffic is global and the same as to all other sites.

This only started in February, the first one I thought a bit unusual, last week's coincidental however now it's getting a bit too regular, heck even on Xmas and New Year's Day this site earned more.

So, it's a more conspicuous than usual example of Google's Data Randomization Program, which I believe is set up to move chunks of our data to hours or even days other than where they belong, thereby increasing the "noise" in the system, keeping us guessing as to what works, and less able to "game" the algorithm in some nefarious way.

So, it's a more conspicuous than usual example of Google's Data Randomization Program, which I believe is set up to move chunks of our data to hours or even days other than where they belong, thereby increasing the "noise" in the system, keeping us guessing as to what works, and less able to "game" the algorithm in some nefarious way.

I don't think it's all intentional, but neither do I think google sees this as a bug. I think it's a consequence of having so many thousands of servers in a distributed processing environment. We tend to forget how stuff is not all in one place, datawise.

Probably a good analogy is how the vote counting system works in a large country. Numbers come in all at different speeds, and you can't safely determine trends without a) sophisticated statistics and informed sampling and b) an understanding of the sources of the numbers.

I think you missed the joke/sarcasm/confuscation there coachm:-)

What I find strange is that for the last 6+ years Wednesdays have been on average one of my best days of the week, this February some of my worst days have been Wednesdays.

Vice versa my weekends used to be my lowest and for the last couple of months or so they have become, not only good, but last weekend my best since 2006.

I'm going to postulate that maybe the consumer market is ahead of the wholesale/retail market buyers and that Joe Public is feeling reasonably confident enough to consider major spending again.

My market is the wholesale manufacture of expensive specialist construction products, not throw away items.

Is there anyone else here in higher-end products like this who has seen a similar trend? And it is a trend since I have seen this data rising since the middle of last year.

I think you missed the joke/sarcasm/confuscation there coachm:-)

I laught at all your posts!:)

...but your answer is this: You don't know. You'll never know. You want to know. You'll keep trying to know. And, you will still never know.

I dunno :-)