Forum Moderators: martinibuster
Anyway, usual disclaimers:-
Mileage may vary, get permission from whoever pays the bill before you use the phone, mobile networks may charge differently from landlines and so on.
In a nutshell, looking at the period of this month to date, with the last 9 complete days being cpm free, impressions are up 1%, clicks up 7% ands ctr up 2% over the previous part of the month with cpm ads.
Nothing stellar there, but the other metrics are quite interesting.
EPC up 35%
eCPM up 33%
Earnings up 34%
These figures are site wide, as the removal is sitewide. I would say that the improvement in some of the smaller channels has been better than the higher traffic pages, but no real pattern to it. I've seen the usual daily ups and downs, but in general I prefer to look at longer term trends. I will keep monitoring of course.
I've known you can remove cpm ads for a while but never bothered. Reason being that the stats show miniscule numbers of them being shown. So the obvious question is what other changes have happened.
Well, I've done NO work on the site the last week or so. I wanted to make no changes whatsoever so as to rule out the effect of changes I've made. There haven't been any "Upgrades" during the period of the test, and I've not seen an improvement in ad targeting. So it's difficult to see what has caused this particular rise in fortunes. However, the fact that dumping site targeted ads coincides with an immediate eCPM rise of 33% can't be dismissed as a coincidence either.
could it be that there are only a few high-paying advertisers
Actually, no!
The site I'm refering to is a fine art reference site that has a dirth of galleries and print/poster advertisers most all of whom I've visited and provide, it would appear, quality products or services. Since this has only been going on for a month I've put it down to some other external factor rather than true competion.
Chapman
but how do you measure "best-paying" if not by eCPM? That's the way to compare performance between contextual and CPM-based ads.
By total revenue?
There's no single "way" to do hardly anything. You're assuming.
How do Google account for cpm ads that are targeted by keyword?
sigh..
Uh. well, they don't account for them, because they don't serve them because they don't offer them?
Or am I mistaken.
in other words, google would have also had a better day yesterday if there were no cpm ads on your site... but why would google hurt it's own earnings by serving up that low-rent cpm trash, when it could have given you contextual ads instead?
i guess that it's possible, given how messed up some of these google algos appear to be, but the very few site-targeted cpm ads that i do get have not hurt my earnings at all.
every site and every situation is different, tho, so it will be interesting to see how the changes that david made pan out over the course of this month.
Anyway, the important question is, regardless of how they measure it, does the algo. sometimes get it wrong...
The algorithm needs to make predictions about contextual ads vs. site-targeted ads. Total revenues vs. eCPM measure pretty much the same thing, don't they?Anyway, the important question is, regardless of how they measure it, does the algo. sometimes get it wrong
I'm not really sure they DO measure the same thing, but I guess it doesn't matter since we don't know what metrics google uses to serve ads, or how they use them.
To answer your questions, whether the algo sometimes gets it wrong, the absolute unequivical answer is that across billions of ad serves, the algo will at least make one mistake.
These threads should be marked "for entertainment value only". All they amount to is that if people want to use trial and error and see what happens, that they can do so (well, if you need a thread like this to tell you that...).
There's simply not enough information about the adserving process to do anything with any efficiency. What works today won't work tomorrow, etc, what works for one page not the other. Cause-effect conclusions are impossible.
Generalizations are impossible. Resistance is futile.
But it's fun to trade theories.
While my numbers are no where near most of you guys... I do notice very frequently 1 cent payouts.Yeah, Google is out to get you ;) People sending you money is not really people taking from you. Unless you feel somehow entitled to more than they give you.My feeling has always been that the 1 cent payouts are taking up valuable real estate on my site - where I could earn 30-70 cents, I just gave that up for 1 cent.
While I believe in trying to serve the best possible ads to my audience...the 1 cent option actually encourages the MFA sites to utilize my traffic for arbitrage.
I will not take from anyone else - but I'll be darned if others take from me.
The flaw in the logic above is you are taking a 1 cent per impression ad and comparing it to a zero cent per impression ad but using it's CPC to compare with. You're comparing apples to motorboats. That one cent payout does not mean you're losing your entitlement, it pays whether it's clicked or not. Try to find an MFA anywhere that pays a penny a view or $10 eCPM.
To answer your questions, whether the algo sometimes gets it wrong, the absolute unequivical answer is that across billions of ad serves, the algo will at least make one mistake.At least one, but if we remember that mistakes do not always mean we become victims we can relax knowing sometimes it works in our favor.
I am thinking that if the SITE targeted ads are appearing then those are replacing the CONTEXTUAL ads which pay higher.
Would it not be true that an advertiser can commit to showing 2000 impressions and the total cost of those impressions are actually higher (in terms of committment) than an advertiser showing a few at lets say 50 cents each?
Therefore I would "assume" that it would be factored into the algo that G should attempt to dispose...(show) those ads as soon as practical.
I would think that this is the scenario where I end up with SITE targeted ads and not the higher paying contextual...
I guess in short what G sees as the total dollar of the committment could end up being more than the POTENTIAL to show a few higher paid ads and therefore shows the SITE targeted ads first.
It seems to me that good business sense from G's perspective could actually mean just that - It makes good business sense for G - but not necessarily for the site owner/publisher. There is noting wrong with that - it just means that the site owner/publisher needs to use good business sense as well.
ARC
I suppose they're better nowadays at matching context and making sure it doesn't lose money for publishers, but it'll be awhile before I go back into the waters, mainly because of the quibble I list below.
Pro
I've used it successfully as an advertiser spending a little under a hundred dollars per day on one site, but obtaining a ton of clicks that averaged to half a cent per click. That site operator may have been earning an extra hundred dollars per day they might not have otherwise made.
A quibble
The concern I have with the CPM program is that instead of showing "advertise on example.com" they display, "advertise with Your Legal Business Name" which although it may or may not be in the whois, I find it awkward in that particular context.
While I am currently inclined to give the site targeting program another chance, I'm holding back because I don't like the way they show my legal business name.
Site targeted ads represent only 2% of overall impressions, have 1/3 of the clicks that contextual ads get, and yet they have 3x higher eCPM than my contextual ads.
I wouldn't opt out. Judge the situation yourselves before making any hasty decisions.
Oh and my stats from past week or so are up as well, 20-30% incrase in daily income with same traffic and I haven't touched the site for 2-3 weeks now.
[edited by: Thez at 12:27 pm (utc) on Oct. 30, 2006]
It's possible that site-targeted ads are such a small part of the overall mix that one or two advertisers can have an impact on site-targeted eCPMs for any given site (even a site with reasonable traffic). Whether this is a problem obviously depends on how many site-targeted ads are being served.
Based on my own travels around the Web, I'd estimate that site-targeted CPM ads are pocket change in the overall scheme of things. Contextual ads are far more prevalent on news and portal sites, mom-and-pop publisher sites, directory sites (including scraper sites), and parked domains. In fact, previous threads on this forum suggest that many publishers--and, presumably, users--have never even seen site-targeted AdSense ads and don't know what they look like.
Site-targeted CPM ads are like AdSense "image ads": They're a sideshow for the main event.
Here is something to think about.
Lets take a small to average site and assume an average of 1000 ad impressions per day (i am being very conservative here). Multiply this by only 2 cents, you get $20 p/day only from site targeted CPM. Multiply this by who knows how many web sites that are running adsense. Lets be very conservative and say that there are only 200,000 sites (over the entire adsense network) that will fit the above description....that’s 200,000 sites * 1000 impressions p/day * $0.02 = $4,000,000 p/day * 30 days = $120,000,000 per month * 3 = $360,000,000 per quarter
Mind you that i have used extremely conservative figures. G gets much more then 2 cents per impression in most cases and there are virtually millions of sites that generate 20 folds the amount of impressions i used in my example.
sideshow for the main event.
[edited by: Web_speed at 11:25 pm (utc) on Oct. 30, 2006]
Lets take a small to average site and assume an average of 1000 ad impressions per day (i am being very conservative here). Multiply this by only 2 cents, you get $20 p/day only from site targeted CPM.
Trying to be polite here, but dja think you might want to take a second look at your calculations above? Methinks there's a...well, numerological impairment, either at your end, or mine!
...and now, perhaps a credibility problem?
numerological impairment, either at your end, or mine!
rbacal, according to my math book 0.02 * 1000 = 20. <whisper politely>...double check yours
Perhaps i should have worded it better:
Lets take a small to average site and assume an average of 1000 ad impressions per day (i am being very conservative here). Multiply this by only 2 cents per impression, you get (G gets...you get a couple of cents) $20 p/day only from site targeted CPM.
[edited by: Web_speed at 11:53 pm (utc) on Oct. 30, 2006]
Since then, my earnings and CTR are up a fair bit. There's no doubt about the increases, although I've made other changes that may also have helped.
The thing that puzzles me about all this is not knowing (and not being able to figure out) which companies' ads were targeting my site. I'd just like to know, if only to be able to remove them from my (full) ad filter.
Whatever the ads were, nobody seemed to want to click on them.
[edited by: Car_Guy at 11:58 pm (utc) on Oct. 30, 2006]
When I checked my reports to see how my site's CPM ads had been performing, those ads had received many impressions, almost no clicks, and almost no earnings.
That's what i am talking about. Try strating a site trageted Adwords campigan. I would love to hear how much you had to pay G just to get a few impressions going.
[edited by: Web_speed at 12:02 am (utc) on Oct. 31, 2006]
rbacal, according to my math book 0.02 * 1000 = 20. <whisper politely>...double check yoursPerhaps i should have worded it better:
Lets take a small to average site and assume an average of 1000 ad impressions per day (i am being very conservative here). Multiply this by only 2 cents per impression, you get (G gets...you get a couple of cents) $20 p/day only from site targeted CPM.
You're basing your "estimate" on a "conservative" figure of someone paying $20.00 CPM for site targeting?
Then you are suggesting that google is cleaning up based on this figure?
Perhaps you are new to the business?
[edited by: martinibuster at 12:38 am (utc) on Oct. 31, 2006]
[edit reason] Play nice, please. Thanks. [/edit]
Whatever the ads were, nobody seemed to want to click on them.
I think the problem with some of these site targeted ads are that they are not all that well targeted.
I recently was hit with an ad for cheap flights to India. It was totally inappropriate for my South American websites.
It started showing up regularly on my 3 sites. The reason I noticed it was that my numbers had fallen and I started checking. It did not show up in the preview tool but there it was on my sites. I can only assume it was a CPM ad.
I figured out the URL and blocked it and the next day my numbers went back up. I had made no other changes.
I think the problem with some of these site targeted ads are that they are not all that well targeted.
Apart from the fact that you might not want your visitors to see badly targeted ads, why else would you care?
$20.00 CPM
Normally I see a handful of site targeted ads running in my ststs with paltry resulting income.
BUT.....
Occasionally they run enough and pay enough to make keeping them as an option very worthwhile.
It seems that the numbers itself of the impressions -no clicks whatsoever- the eCPM and the amounts paid tend to be funny:
Now loosing say a 100 impressions of CPC ads that run at a eCPM of $4.5 cost me $0.45, but replacing them with 100 CPM ads at $6 eCPM earns me $0.6 ... so I still win (as GOOG says you do).
Yet disabling these is supposed to increase my revenue by 30%? Sure they didn't also clear the smartpricing?
If disabling CPC disables smartpricing, increases a minimal bid, or whatever other good as side-effect, now that would be something to jump for. But so far (aside from the single digit impression days) the math does not support disabling site targeted ads yet. Unless it has one of those side effects ...
Normally I see a handful of site targeted ads running in my ststs with paltry resulting income.
BUT.....
Occasionally they run enough and pay enough to make keeping them as an option very worthwhile.
That's been my experience, too.
As for the earlier comment about site-targeted ads not always being correctly targeted, it's important to remember that they aren't supposed to be contextual. It's up to advertisers to decide whether (to use a hypothetical example) they should be running ads for boxes of chocolates on sites for pastry buffs. (The advertisers might figure "damn right," on the theory that pastry buffs are likely to be the kind of people who nibble bonbons when they aren't devouring eclairs.)
In this respect, site-targeted CPM text ads are more likely display ads than AdSense's typical CPC contextual ads. I've seen geotargeted Midwestern "museum weekend package" display skyscrapers on a European travel site, presumably on the theory that Midwesterners who like and can afford European travel are the kind of people who'd enjoy driving to [Midwestern city] for a weekend of museum-going, dining, and shopping. Seems jarring at first, but it makes sense when you stop to think about it.
$20.00 CPM
For all we know...it might at times be even much more (for G that is, not the publishers). Can you really tell how many site targeted ads appeared in that one ad panel impression? You see 4-5 ads...how many of them are paying CPM site targeted rates and how many of them pay per click or perhaps both? Heh?
One single ad impression can easily net 2 cents (if not more) in G's pockets...and believe it or not, they do!....check G’s latest inflated (surprise* surprise*) earning figures.
[edited by: Web_speed at 4:27 am (utc) on Oct. 31, 2006]
For all we know...it might at times be even much more (for G that is, not the publishers).
For all we know, the earth could be flat, the moon could be made of cheese, and Mars could be home to little green men.
