eCPM long term trends

How are yours?

[zett]

So, just out of curiosity I ran a 200 day moving average on the eCPM value, across the whole data set (2.5 years of daily data). And I put that in the same chart with the 30 day moving average. With very interesting results:

All this year, the 200 day line shows a steady downward trend. Like the altitude on a plane run by autopilot during approach. Steady. Decline.

But here comes the shocker: I noticed four situations where the 30 day average line was about to cross (break through) the 200 day line from below. BUT IT DID NOT HAPPEN. Instead, the 30 day value "rebounced" at the "virtual roof" of the 200 day line.

No wonder that the eCPM continued its steady decline.

Then I thought about possible reasons. Clearly, I think it is a money grab from GOOG. And then I asked myself - what would I do to achieve a steady decline that can not be spotted as such? The answer: find a mechanism that produces almost random daily values, based upon long-term averages, e.g. 30 day line and 200 day line. Pretty much like the stuff stock brokers use for their work.

So, to me it seems really like there is a target eCPM value to be met by a certain date. Adsense reduces the total payout to meet this target, no matter what actually happens onsite (traffic, conversion rate, ad bids). There is no other explanation to me for the weird eCPM behaviour (one day it's $X.XX, next day it's $X.XX/3).

I suggest that you pull up the same averages (30 day, 200 day) on your data set. Of course, I would be really interested in your analysis of the data.

Having said that, my standard disclaimer applies: Results for my sites only, I know I should not question Google Adsense, I know it could be all just a matter of supply and demand. Please do only answer if you want to contribute your analysis to this thread.

[frakilk]

This is very interesting zett, I had thought about this also. What strikes me is for all the up days and down days you still never break above a certain level. If I were Google and I wanted to mask a gradual decline in eCPM I would do exactly what they seem to be doing i.e.
- Give the publisher a few down days. Cue publisher ranting and raving about crappy eCPM
- Then give the publisher a few up days (or course never raising above your allotted eCPM average level, in fact it will fall slightly short to keep the decline going). Cue the publisher feeling relieved and wondering why he/she ever questioned it in the first place.

Of course this tactic would fall apart if a publisher does a year comparison. But then of course other factors can be proposed as the cause e.g. economic downturn.

Yes it's a conspiracy theory but sorry my trust has been broken.

[cornwall]

I was interested in your theory, and ran those two moving averages myself.

Interestingly my experience is the converse, with recent pick ups in revenue, my 30 day moving average over the last few weeks is above the "ceiling" that you think might be there.

My view still is that Adsense is a "zero sum" game. There are winners and losers, but few can expect to be a winner all the time (similarly one would not expect to be a loser all the time!)

So you can expect to break the 200 day line in the graph some day (perhaps) ;)

[zett]

frakilk & cornwall:

Thank you for your replies and the input.

It's good to see that I may be an exception with my findings. That gives me hope. Definitely. However, I will dig deeper into this issue and look at the combination of various averages (i.e. EPC vs eCPM) to get a clearer picture.