Forum Moderators: martinibuster
Keeping that in mind, I recently changed a 300*250 box (which mostly held 3-4 ads, but also got a lot of site-targetted impressions) to a 234*60 slot and moved it from about halfway (two scrolls) down a page up to borderline above/below the fold. Through css I posisitoned it so that it is the first adsense block on the page to be filled when the page loads.
What I expected to happen was that overall eCPM would increase as I'd be showing only one advert, (the highest paying one as it would be the very first to load) in a far better located spot according to the heatmap. I assumed that the number of impressions of site-targetted adverts would fall, as I'd be showing just one advert rather than 3-4, so I figured the pool of CPC adverts would "last longer".
The results were not however what I expected.
Instead, site targetted impressions rose -- by over 250%, while contextual impressions fell by half. Both change smeasured Vs the historic, 30 day moving average.
Perhaps that would be understandable if the site-targetted eCPM rose, but it didn't. It's actually fallen -- fallen by 50%. Contextual eCPM has increased (more than doubled) -- which is what I expected, but this increase has been more than offset by the drop in the overall impressions.
The net effect is I'm earning less for the channel in question.
So, what I don't understand is why my contextual impressions (which have a far higher historic eCPM) fell by half while the site-targetted impressions have more than doubled.
Nothing else on the site has changed -- all the text is identical before and after.
Any pointers suggested the mistake in my expectations would be much appreciated.
Regarding my initial impression of how site-targeting performs in the "auction", here it is from the horse's mouth:
"Site targeted ads compete in the same auction as contextually targeted ads. They will only appear on your site if they represent a higher eCPM than any combination of contextually targeted ads competing to display in the same ad unit. The AdWords auction process takes into account factors such as bid price and clickthrough rate in calculating Ad Rank, the ranking system determining which ads will appear on your page. Opting into site targeted ads should only improve your overall revenue potential by deepening the pool of advertisements competing to display on your sites."
[adsense.blogspot.com...]
Really doesn't seem to be the case in my experience.
Anyway the say it should work is that Google will show the site-targeted ad if you're projected to earn more overall. Since you reduced the block from 4 ads to 1 ad, your anticipated earnings from that block are a lot less, hence site-targeted advertisers need to bid less now. I think that's the way it would work.
Site targeting only works if your site is about a very narrow topic. If you have multiple subjects and most pages are different topics...it simply doesn't work as it is not "page-targeted".
Site-targeted ads are being bought by the impression, not by the click, no whether the ads "work" or not shouldn't have an effect on earnings. If National Widget Co. wants to buy "Your One-Stop Widget Source" ads on your pages about widgets, whatsits, and thingamabobs (possibly on the theory that whatsit and thingamabob buyers are also widget buyers), why look a gift horse in the mouth?
Site-targeted ads are being bought by the impression, not by the click, no whether the ads "work" or not shouldn't have an effect on earnings.
EFV I am quite aware of how site targeting works (there are many other members besides yourself that do know these things).
I'm still trying to figure out what your post had to do with my post... My post simply stated it works best if your site is about a single topic. If you have something to say in agreement/disagreement then post.
I stated it does not work well for sites that cover a broad range of topics. For example if Wikipedia ran ads, site targeting ads would be irrelevent on most pages, same as if you have a site with broad content. Say I visited a site that covered travel in Australia and Canada. What good would an ad do about Quebec on page about New South Wales simply because someone targeted it for Quebec, Canada.
If you replace a relevent content matched advertisement with an off-topic site targeted ad, it has everything to do with earnings.
[edited by: The_Contractor at 9:11 pm (utc) on Oct. 4, 2007]
Site-targeted ads are being bought by the impression, not by the click, so whether the ads "work" or not shouldn't have an effect on earnings.
Obviously that's not been my experience.
Anyways have mailed Adsense asking to turn it off and I'll see what happens. It's a bit frustrating as I know I've got some pages (forum, pages on obscure locations) where site-targeting is good - it's a shame there's no method to specify in the adblock if you want to allow site-targetting or not, instead of having to make a site-wide judgement.
I'll report back with the results once it has been nixed - thanks for the feedback all.
The Contractor, "site-targeted" ads are as old as advertising, and they were being used in general-interest media long before the Internet and PPC ads were invented. Just look at the ads in TIME or the main section of THE WASHINGTON POST.
Again, you don't need to give an explaination where non-contextual advertising came from (we are aware of this)... some of us are much more familiar with the online world than those who came from print media...the user simply asked why this impacted his/her earnings. I simply answered without a lesson/history in advertising...why does it always have to go off track...hehe
After years of running AdSense CPM ads, I think this product is a dud for my site (many related topics if you're wondering), I also happen to think that it does not work as advertised in terms of competing for the ad spot and raising the site's overall earnings, observation and numbers prove otherwise, at the very least even if it is as advertised, it's zero targeting $crews up your ctr epc ecpm and earnings.
And yes, just to avoid having to come back and post it later: YMMV!
And yes, just to avoid having to come back and post it later: YMMV!
That's something that people often tend to forget around here. Site-targeted CPM ads work great for some sites and badly for others. But even on sites where contextual ads typically bring in higher earnings, site-targeted CPM ads can be (or may be) useful "filler ads" on types of content that have low clickthrough rates.
But even on sites where contextual ads typically bring in higher earnings, site-targeted CPM ads can be (or may be) useful "filler ads" on types of content that have low clickthrough rates.
Sure, that May in some cases be true, but given you can't disallow site-targetted adverts from some sections of your site (and so leave them to play their role as "fill"), you're left with an all or none choice, which is less than ideal.
Site-targeted ads are being bought by the impression, not by the click, no whether the ads "work" or not shouldn't have an effect on earnings
My understanding of site-targetted advertisements was that they would only appear when they were the highest paying form of ad for a slot on a page
I wonder if site targeted ads also appear based on ctr history as contextual ppc ads do. (That is if the site targeted 'per click' type do exist within Adsense)
Does anyone know for sure?
[edited by: Scurramunga at 1:29 am (utc) on Oct. 5, 2007]
I thought that there were two types: 'impression' and 'per click'.
Site-targeted contextual ads are in the works, according to Google, but I don't believe they've been implemented (except, presumably, in beta).
If they allowed us to specify CPM ads by channel I could make good money. But if they did, the web will be full of pages with CPM ad units hiding at the bottom doing nothing but incrementing like site traffic counters.
Hobbs, what could possibly make you so cynical about the human race? :-)
Overall earnings remain about the same (admitedly only on a 7-day moving average), but not all that surprisingly, what has changed is the number of clicks, which has increased by around 20%(on the same moving average).
What I'm hoping is that as a result of the increased clickthroughs, they'll be a rising conversion rate and eventually I'll see a rising payout per click -- but at the end of the day there's a huge swag of guesswork worked into all this!
If nothing else, it's a better user experience in that at least my readers are being shown adverts that they want to click on rather then generic round-the-world air tickets that I used to see.