Forum Moderators: martinibuster
About a month ago we started a series of tests that at first looked very promising. When we converted a majority of our pages to the new format all of a sudden our eCPM took a plunge. After about a week we completely reverted back to the original formats and positioning to try to recover. That was two weeks ago and our eCPM is still in the dumper.
Today we realized that our testing inadvertantly put 3 regular ad units on a page plus a referral link, which is also considered and ad unit, so we had a total of 4 ad units on many of our pages.
We have enough pageviews that our eCPM has been very consistant for at least half a year, so smartpricing has never really been an issue with us before. Is there a chance that we have received some sort of penalty for having too many ad units?
You can place up to three ad units per page (in addition to one link unit and one referral button per product).
When you ran the split test did you notice any differences in ad targeting? If your "B" page had more ad units might it have resulted in less targeted ads? Or maybe the 3rd ad unit was inadvertently showing in a more prominent place resulting in lower cost clicks?
By the way, what were you testing? Above and below the scroll; left-hand, right-hand; ad size; number of ad units? This might turn out to be an interesting study in ad placement.
The referral link doesn't count as a fourth ad unit.
Oh... I read the TOS wrong. I guess that's not the issue.
On my main "content" pages I have had a 160x600 on the upper-left where web site navigation is normally found, and a 468x60 at the bottom of the content. This has consistantly produced good results.
In this test, I placed a 300x250 box at the top, right of center within the content of the page. Initially 10% had the 300x250 and the 468x60, 10% had the 300x250, the 160x600, and the 468x60, and 80% had the original 160x600 and 468x60.
My initial findings showed that having all 3 ads increased increased page CTR substantially, and eCPM went up about $5.00, so we converted 80% of the pages to use all 3 ads. After a few days eCPM started to drop like a stone. Trying not to panic we waited a few more days until we decided to just go back to the original layout of one 160x600 and one 468x60. Now eCPM is $4.00 less than the norm, which is a enough of a drop to have to eliminate 2 part time employees if it stays at this level.
I'm not sure what to make of it...
As a first step, do you have or can you set up channels for each of the ad positions? This would at least tell you how each individual block is performing and you might be able to make some guesses from that.
ecpm is totally worthless for troubleshooting a website... you mentioned ctr for a second there, but veered away from it :-/
was it the ctr that caused the ecpm to drop like a stone after a few days?
It's better to concentrate on EPC than eCPM I think.
In the last week my EPC has been all over the place. A few days ago I improved some ad positioning on a small portion of my site and it seems to have increased my CTR. But even CTR can be all over the place for no obvious reason.
For a million plus pages it's hard to do, but have you eyeballed the targeting at all? Is it in the ballpark?
It almost seems like you might have gotten caught up in a confluence of events. Your layout change, maybe a change in ads/bids on the Adwords side, possibly some poor targeting possibly caused by one or both of the first two, possibly a smart pricing adjustment. Who the heck knows.
The only thing I can think is that if you've put everything back to the way it was before the test, and if it was a smart pricing adjustment, and if everything stayed the same on the advertiser side (bids, inventory), then you might see a recovery as smart pricing is readjusted over the next couple of weeks. (And that's about as wishy-washy a statement as there can be.)
BTW, Don't confuse per-ad CTR with page-wide CTR. Increasing the number of ads on a page will invariably decrease the avg CTR since they can only click on one ad per PAGE (vs ad) impression, tripling the ads, will divide the CTR by 3 when looking at either ad or page CTR. Also the added ads will generally be lower EPC than the first set further lowering the avg EPC. CPM is more help in this case since this is what GAd strives to optimize anyway to the benefit or detriment of all else.
Also BTW, I don't think there is any actual penalty for too many ads either, although it is expressly discouraged by GAd in their policy. One publisher example recently flaunted by GAd and even linked from their own examples page as an exemplary website showed 4 quite obvious ad units on one rather long page.
my top-earning page appears to have that happen quite a bit, because the web surfer apparently wants to keep the content of the page open(?)... the ctr on that page is really good.
your comment about overall daily site ad impressions changing things is very interesting, i wonder if that could have been a factor in my epc taking a beating back in december... it happened after a spike in traffic.
