Experiment to Increase EPC - total failure

[sailorjwd]

You may have read that a week or so ago I experimented with removing some Google ads and replacing them with ads advertising my own high value pages on the website.

I placed a strip of links on the far right side pointing to these high value pages. I didn't wrap the google ignore code around this strip.

I initially announnced an immediate increase in average click value - it was dramatic rise of about 30 to 40%. This lasted for a few days but gradually started to trail off.

Now two weeks later my CTR is at an all time low and EPC is about the same low value of 4 weeks ago.

I had some time today to look at many pages and what I found was similar ads on ALL pages. The ads were for the high value pages but I wasn't getting any clicks since they didn't match the content on each page. Googles algo saw a couple of keywords on that ad strip and it totally polluted the topic of the page in G's eyes and allowed the high-bidding off topic ads to show up.

To try to repair the damage and keep the original good effect I've changed the ad strip to an image link of it without alt tag. Therefore there are no off topic words on the page. I hope google can't read images yet.

I suspect also that the epc went down again because I got smart priced severely.

I hope everthing bounces back as G reindexes the pages.

[wyweb]

[I hope everthing bounces back as G reindexes the pages

I hope it works out for you too sailor.

I didn't wrap the google ignore code around this strip

Would have been interesting to see the results if you had. Many publishers have had problems with section targeting and the ignore parameter. That hasn't been my case at all. I've used both successfully. Especially the ignore. In my case it works as it's supposed to and as google claims it should. Many apparently have not had this luck.

I use "alternate_ad_url" to display links to interior pages of the same website. My niche is large enough that ad supply is never really an issue but if there ever is an issue the alternate ads basically mimic site navigation. I ran some click tracking once on one of my larger sites and the alternate ads weren't getting hit at all except by bots. There was the odd click every now and then by a bona fide visitor.. javascript disabled most likely. On leaderboards and the big skyscrapers, the 160x600 ads I use some affiliate graphics for alternate ads but they rarely get hit either. I tracked them as well for a time but it got boring so I quit.

Google's algos take time to actually settle in and get to know a site, if that makes any sense. Traffic volume seems to be a factor here. I have a small site that gets maybe 10 - 15 unique a day, maybe 30 page views total daily. It's been online for over a year now and targeting STILL sucks from time to time. My biggest site gets 4K unique a day and targeting is almost immediate and always accurate. What distinguishes these sites? Basically just traffic volume.

I suspect also that the epc went down again because I got smart priced severely...

Yeah well, welcome to my world....

[surfer67]

I had the exact same experience this month. For years I had only been running one 468x60 ad at the top of every page, so last month I decided to also add a 368x280 ad size to a number of pages. Initially, the ctr and epc went up 30% for about a month. This month the ctr has crashed and I'm earning less than I did before. For me the problem appears to be the ctr, which has been cut in half. How can ctr go down 50% after adding a large 368x280 on every page?

[sailorjwd]

Surfer,

Note that i wasn't adding Adsense ads. I removed a few actually. the point is I was adding internal links to high epc pages.

[vordmeister]

I wonder what the algo behaviour is in this case.

It's fair that external links should influence their ad serving algorithm, but it's weird that links would dominate it above what the page is about.

Others on here have suggested the algorithm is scared of change. Is it possible that they see a change and think "whoa there's something different here, we'll need to think on this" and default ad serving to the generic theme of the website?

Why they have to think for so long might be down to their anti-spam stuff. They wouldn't want to instantly serve the pay day loans ads to some site that is trying to play the system.

For the rest of us it just means we can't change anything without massive loss of earnings for us and GG.

[sailorjwd]

I had a similar problem early on with adsense (2005). I'd have a page about topic A - a few 100 words. If I had a single mention of topic B then all the ads would switch to B since B was bidding very high. I went so far as to put the special word as a graphic so G couldn't see it.

The problem seemed to go away to a large extend for internal links that were on a left/right border especially with the ignore code.

but its better now in general with an image link since I'm not screwing up the keyword density of the page either. (not that i'd ever intentionally adjust keyword density :)

[purplecape]

surfer, you said:

"How can ctr go down 50% after adding a large 368x280 on every page?"

Easily. You doubled your ad impressions, right? If you got no additional clicks, just split the clicks you were getting, your CTR would be cut in half.

[surfer67]

Easily. You doubled your ad impressions, right? If you got no additional clicks, just split the clicks you were getting, your CTR would be cut in half.

No because it's the page ctr, not total ad impressions.

[IanCP]

the experiment makes sense to me

BUT

As I regularly say:

"Change Nothing"

FWIW< YMMV

[vordmeister]

Just had a look at my own stats - they had their biggest weekly drop ever between the week before last and last week. About 33% lower earnings.

I didn't make any changes but seem to have replicated your findings. So it could be co-incidence rather than an effect caused by something you have done.

[photojack]

I agree with Vordmeister 100%. It definitely could be a coincidence. I have had a similar drop in earnings during the last two weeks, although I haven't changed much anything.

I've been doing this for years, but this is the first time I encounter such an explained and significant drop in earnings.

[sailorjwd]

The ads are starting to return back to the main topic of the various pages. At least now people will see ads related to what they were searching for.. whether they click or not will be up to them.

[purplecape]

No because it's the page ctr

thanks for clarifying. As others have posted, maybe outside factors were involved. (I've also seen a drop in CTR recently, and have done nothing to my site)

[londrum]

How can ctr go down 50% after adding a large 368x280 on every page?

if you're adding another four links to a page that's already got four or five at the top then maybe those extra ones aren't all that great. you might be using up all the good ads at the top, and get lower quality ones down below.
if people's eyes are drawn to the new block more than the old one, then maybe that explains the drop in CTR.