Your question is similar to the situation with a drop in CTR (here was
my response to that:
http://groups.google.com/g/de77e6bf/t/771ecf20c3518951/d/1047471e2012f58f)
CTR and eCPM are very related. They are both measures of ad
efficiency. Similar to that post, if your requests/impressions are
below about 100,000 per day, then you can expect very volatile
results. A few days is not enough data to pinpoint the source of
dramatic changes.
However, if your CTR has not changed, then a drop in eCPM is because
of a drop in the value per click. eCPM is "dollars per 1000
impressions." It is an efficiency measure (like miles per gallon or km
per liter; CTR is another efficiency measure - clicks/impression). The
US Independence Day holiday (July 4th) resulted in higher US usage for
most apps. Higher usage drives down the price paid per click (it is
easier for advertisers to get clicks with more usage, so they do not
need to pay as much for the same CTR).
eCPM is equally sensitive to both CTR and $/click. If your eCPM
dropped 80%-90%, then your $/click would have to do the same if your
CTR is constant. We saw about a 30% drop in $/click - which we
expected (actually, we expected 40% - glad AdMob didn't meet that
expectation!). If your $/click dropped 50% and impressions increased
"insignificantly" by 20% (assuming a corresponding decrease in CTR),
you would go from $0.80 eCPM to $0.33. Not sure how you could land at
$0.10.
What are your last few weekends like? Did your usage change
geographically (higher usage outside the US could offset a drop in the
US, but possibly result in lower $/click). Also, do you filter or use
location-based ads? That can be effective at getting clicks, but
result in a drop in pay for a holiday weekend. If so, hold steady - a
good CTR means your users like the ads, you just need advertiser
budgets to recover from over-usage this past weekend. Or, try
different filters to see if less relevant ads (lower CTR) bring in
more overall revenue.
When you try new filters, set up a spreadsheet and record the number
of impressions, clicks and pending revenue several hours before the
change. Then record it again at the time of the change. And again
several hours later. You can calculate CTR, eCPM and $/click from that
data. Then change them again, and record. It is harder to do that now
because some Adsense revenue does not even appear on the Pending
revenue, but if you do it the same day or at the same time of day
during similar days of the week, you will see the impact of the
changes. The aggregate numbers (late night or early morning activity)
mask the real-time effects of the change - especially the closer to 12
GMT you get. You have to sort out the effects of the previous hours.
(You also want to try to do it at times you have steady usage -
variations in usage can also be from variations in user demographics,
which can be sensitive to time of day measures.)
We do this constantly to make sure we understand the impact of
filters, design changes, time of day, day of week, etc. effects.
You probably want an answer from Google that accounts for your
particular app(s) performance - but your question is too broad for a
good answer from them. They might be able to explain an overall
decline in $/click or increase in overall app usage (which would be
great if they could give us some kind of benchmark). But then it would
be hard for us as developers to see something like "clicks average
$0.10 today" and then realize some of us only got $0.05. We can't all
be above average, but I'm sure we all want to be.
How's it looking today? We're holding steady, better than the weekend,
but lower than last week. Not that surprising.
-Jim