I think mbennie is saying eCPM may not be a good way of measuring fluctuations if you haven't truly looked at your traffic
Exactly. Lots of posters here are complaining about eCPM dropping. Evidently bots can negatively affect eCPM.
Blocking is a dangerous game, one could argue that restricting site traffic to only Europe and North America could raise your EPC and eCPM, but in the end of the day your grocery does not get paid in eCPM & EPC but in real money, so the impact on the overall earnings is what needs to be taken into account.
We didn't just whack continents - didn't do anything geographically. We identified known bots by UA and suspect bots by IP. Then 403'd them all.
We didn't do it to increase eCPM. We did it to reduce server load and prevent scrapers from stealing content.
I was surprised by the difference in AdSense stats and thought some users here might appreciate another way to look at the problem of declining eCPM.