Forum Moderators: martinibuster
eCPM
The main factor used by the Google Algo is probably eCPM: ads are assigned in descending order by the eCPM achieved, historically, on your site. This is to give you, the publisher, ‘the best overall return’.
Eg: suppose you have a single skyscraper on a page, with 4 ad slots. Starting with the the first (best) ad slot, the Googlebot gives it to the ad that, according to the ad history on your site, yields the highest eCPM (not the highest bid, note). Then it does the same for the 2nd slot, and ditto with the 3rd and 4th slots.
How an individual ad's eCPM is calculated
The ads’ eCPMs can be calculated as the sum of all the click values, times 1000, divided by the number of impressions. But this can also be viewed, more usefully, as: $AverageEPC x CTR x 10
However, the CTR used is not the CTR you see on your report: you see the CTR for the 4 ad slots put together; but Google keeps and uses the CTR for each individual ad. Therefore, of those 4 ads, the top slot CTR is probably much higher than the bottom slot.
Eg: suppose you have 4 advertisers (competing for space in the single skyscraper), each bidding 50c, and your reports show a click through rate of 8% for that page. The four ads might have a CTR of 4%, 2%, 1% and 1% each, which means that the top ad achieves an eCPM of $20 (.5 x 4 x 10) and the bottom ad achieves an eCPM of $5 (.5 x 1 x 10).
How an MFA gets in with a high eCPM
Then, along comes an MFA with a low, discounted bid (say, 10c, discounted because of a low conversion rate). As it is a new ad (to you), without any track record on your site, the Googlebot takes the CTR from its’ search and other sites (say, 7%, because MFAs use clever, sometimes dishonest, wording to get high CTRs). This means it has an eCPM of $7 (.1 x 7 x 10).
Therefore, although the MFA is bidding much lower than your other advertisers, it wins the third slot because the ad’s individual eCPM is higher than two of your advertisers.
How MFAs spread
Once it is appearing on your site, and achieving a high CTR, it suppresses the smartprice on that page and your site (because it has low conversions) causing the value of your top advertisers’ clicks to drop. Gradually the MFA creeps up the page and other MFAs start to invade the page as well.
The solution
The way to prevent this happening, in this example, is to reduce the number of ads on the page to two and price the MFA off your site. You could use, say, a vertical banner instead of the skyscraper and putting it in a hotspot to keep the CTR high.
Eg: if you maintain an overall 8% on the page, the top ad might have a CTR of 5% and the bottom one 3%, giving eCPMs of $25 and $15. This puts the price of the ad slots out of reach of the MFAs.
I think the calculations involved are actually much more complicated than I've explained. But hopefully it illustrates the general principle and the benefit of reducing the number of ad slots when the average EPC is low (by comparison with adwords rates).
I hope this is helpful. And in your battle against MFAs, may the force be with you
Google is simply going to have to start banning MFA domains... The problem is, they are addicted to revenue growth and Wall Street expects it.
Scott, I suspect Google want to be able to get rid of MFAs. Although it is sometimes said that MFAs make money for Google, imho they reduce Google's earnings.
For example, it might be said that some MFAs buy a lot of Adwords space. But this isn't new money coming into the system, their Adwords expenditure is financed out of Adsense earnings. So what they 'put in' to the pot was already there in the pot beforehand. In fact, the net effect is that they are taking profit which otherwise would be shared between Google and original content/functionality providers.
Another financial argument is that the presence of MFAs creates more competition and thereby drive prices up. But the evidence suggests the contrary, they drive prices down. That is the experience of Adsense publishers, and in fact the only way MFAs can succeed (by being able to buy cheap clicks in the first place).
In addition, MFAs put off potential Adwords advertisers, as is demonstrated by your own comments about reluctance to invest in the Content network.
It seems to me that the financial impact of MFAs on Google is a negative one. Their motivation, whether a desire to "do the right thing" or "satisfy Wall Street", should lead them to adopt a strategy that deals with MFAs.
Imho, the problem is not one of "whether", but of "how". Whatever tactics Google come up with MFA creators will try to find a way around and/or there are risks of unfairly penalising bona fide publishers. Smartpricing is perhaps a good example: it was probably introduced, in part, to combat MFAs, but now MFAs are using it to their advantage.
I reckon, therefore, that if someone can come up with a practical, reliable and fair way of combating MFAs Google would jump at it.
However, publishers can play their part as well. I've started this thread (and others) in the hope that publishers will adopt strategies that price MFAs off their websites.
However, that doesn't stop the problem of MFA's appearing all over Google's own search pages. That is a huge problem in itself that only Google can resolve if it felt minded to.
========================================
Google sent me an Email asking me why I was not advertising on the Content side... I told them I was not going to use the Content side of the system because of Click Fraud and the MFA sites. Google stopped trying to get me to use the Content side of the system.
First (very early) stats in from my test.
Pages with the VerticalBanner 120x240 (max 2 ad slots) have an overall *slightly* higher AdSense CTR than any other tower format including the usual-best wide skyscraper (160x600).
The test has been run on ~5% of page views over the last 24 hours or so, with the banner in different positions (left, right and right-dropped).
This may just be a temporary novelty factor, of course!
Rgds
Damon
I would 'guess' that G are quite happy for MFA to burn and churn their cash away on ads but they would obviously like them to burn the ads out asap and take as much money as possible from the advetisers in the process.
i.e a ad may have high ctr - gets lots of clicks initially but then with landing page quality etc taken into account then these low quality destinations have their ad prices raised and are forced out (at least temporarily). Overall - if the system works in this way quality should always win out and the publisher and G will always win because of the ability to invite MFA's into the system and reject them by a process of we'll take your money while we do some quick iterations to find out the real quality of your ads.
Of course - the system does not yet work like this due to numerous factors, lack of real intelligence and tracking, algo limitations(?) people overusing ad filters etc (I am saying this last bit with my tongue in my cheek by the way).
But this isn't new money coming into the system, their Adwords expenditure is financed out of Adsense earnings. So what they 'put in' to the pot was already there in the pot beforehand. In fact, the net effect is that they are taking profit which otherwise would be shared between Google and original content/functionality providers.
Google makes more money with more clicks. Period. The only ones who lose out are the original high quality content providers. Period.
The people with the original money is the advertiser with some widget to sell. Let's say they are going to pay $1 for a click.
Let's assume that Google pays 70% to the publisher and keeps 30%. Just an assumption, the percentages don't matter. Let's also assume that number is fixed.
So if Publisher gets and ad for a widget and Visitor clicks on that ad, Publisher gets 70 cents and Google gets 30 cents.
That's the straightforward case.
Let's now assume that the MFA gets a 50% CTR and they always go to high paying ads of Advertiser. So if MFA gets his ad on Publisher's site and gets two clicks and pays 10 cents per click, Google gets 14 cents and Publisher gets 6 cents. MFA gets 70 cents for the click and Google gets 30 cents.
So while "only" 1 dollar has come into the system the distribution has changed.
Publisher gets 6 cents, Google gets 44 cents and MFA keeps 50 cents (70 cents minus 20 cents for his ads).
That's how arbitrage works, although the numbers are totally hypothetical. Google has a huge incentive to keep MFAs, they return a huge amount of money back from white hats to Google.
Ok?
EDIT: PS, I'm not saying there aren't OTHER reasons that Google should consider when deciding to get rid of MFAs, or not. I'm just saying there is a strong short term financial incentive to keep them around.
[edited by: RonS at 11:51 pm (utc) on June 7, 2006]
You can only skirt money out of the golden goose. MFAs do that then leave for the next Web money grab.
A simple rule would be to ban any landing page with Google Ads. Sure they are other way around that for a spammer, but it would be more work.
Then perhaps, ban any landing site with Google Ads - but of course, this is sensitive as real honest publishers and advertisers will be affected by this.
By the way, I found Jomax's comment about not sharing secrets interesting. Jomax, does it hurt your bottom line when people know how to combat MFA better? I'm, you know, just asking and not making any judgement...
;)
in regard to how massively mfas harm publishers i just want to add an example:
by now i know my stats very well, just like "my" advertisers. if a new advertiser comes into the game and shows noticable often in my ad blocks, say every second page view, i start to investigate. few days ago, by means of a single advertiser, i had the opportunity to analyze an interesting case. important to mention, you can't categorize him as common mfa. he only has some of the same obvious "qualities" which enable him to play the same arbitrage game. what did he do?
- he anticipates the potential needs of my audience by offering an additional service
- he adjusts his ad copy perfectly well to my specific content
- he offers a catchy incentive
what's his profit?
- by offering this service, he's filling his site with user generated content
- he has adsense ads on his page
- he offers a variety of additional menu points
perfect, one might say, an ideal of an advertiser, well done! i send him a steady stream of clicks and he pays me a steady stream of adsense cash in my pocket. somebody to keep warmhearted in an ad block, a symbiosis, a win-win-situation!
sadly, that's wrong. in fact, he's hurting my bottom line and i lose money on him. why?
- just like mfas he does arbitrage. since he sells no tangible goods he's very much forced to pay low to keep a profit. i expect him to pay minimum cents. it immediately showed in my stats. typical effect: ctr up, epc down.
- he shows a very high degree of professionality, he obviously bids on a large variety of low budget keywords (no site targeting)
- in the following days, smart pricing kicked in, thereby lowering my epc once again considerably. why is that? well, i suppose there are some kinds of tricks to show low conversion
ok, his type of business enables him to play that game. intelligent advertisers pay less. can i blame him for his strategy? certainly not.
but as publisher there is no way to cope with that double penalty of minimum payout plus smart pricing. so i watch the game still for a few days and then i'm forced to kick him out.
what do we learn from this case?
- ctr and epc are conflictive measures
- kick out every advertiser who has nothing to sell
- even or more than ever "good" advertisers with "well going" ads can hurt publishers
- caution of too attractive ad copy. investigate.
- symbiotic advertisers can turn out to be competitors
actually, i'm sick of doing the job cleaning up the dirt for google. i want an algorithm that provides the highest possible payout for my sites and not the highest possible click count. i want a smart pricing algorithm which doesn't work in favor of arbitrage businesses and against honest publishers. i want google to realize, that market forces will just not catch on finally in this imperfect market.
i suppose there are some kinds of tricks to show low conversion
This is one of the concepts that the whole "MFA'ers forcing us into smartpricing" seems to revolve around. Admittedly, I think they're smarter than we give them credit for, and have figured out every way to game the system possible. But as an AdWords advertiser, I'm just not sure about them being able to tweak the conversation stats.
From AdWords "How are conversions determined?"
A conversion is registered when an ad click leads to an event that you consider valuable. Depending on the business, a conversion can be defined as:- A purchase
- A signup or registration
- A request for more information
- A page view
- A demo download / game playThe page where you confirm that a user has successfully taken one of these actions is the conversion page. The Google Site Stats text block is placed (generally, the Thank you for your purchase/subscription/visit page), and is then seen by the user after a conversion occurs.
I only dabble a little bit in AdWords to drive some traffic to my new site. Does anyone know about this "Google Site Stats text block?" and ways we might be able to hunt for it on MFA sites? If we could find a correlation between MFA sites actually using this indicator - that would lead me to believe that they're gaming the conversion numbers to benefit themselves. Perhaps Google would actually listen to that!
(I know, I know - can't believe I actually said that... like that'd ever happen!)
On the search network, though, traffic going via MFAs is a straight loss for Google. Rather than getting the whole $1 themselves, they get 10c for each of the clicks bought by the MFA and then 30c commission on the final ad - the final distribution is 50c to the MFA and 50c to Google (ie Google's loss is 50c).
So, if Google were really cynical, they would ban MFAs from the search network but let them on the content network, that would maximise their income.
Sounds as if you're cutting off your nose in spite of your face huh?
(.. FYI.. I am not an MFA site owner. I'm just playing devils advocate here. But seriously, why care?)
Because many MFA's write deceptive ad copy titles like:
Click here for the latest info on widgets!
Your 1 widget resource!
Hundreds of bargains and deals for Widgets!
Your visitor clicks on one of these ads and is invariably confronted with a page featuring two 336 rectangle ad blocks side by side, and a leaderboard underneath that.
Unless the visitor is a click monkey he/she will be annoyed that the link recommended on your site lead them to a spam page with adsense.
You just lost the trust of that visitor for a 3-5 cent click.
That example may seem counter-intuitive to you, but the reason is that MFAs don't actually 'pay enough' to be on our sites.
Google's stated position is that ads aren't placed on sites according to what they pay, they use a formula called 'Adrank', which is 'quality score' multiplied by 'max CPM'. So, if MFAs can manipulate their quality score to be very high, they can pay very little to get onto your site.
We don't know exactly how that quality score is made up, but Google cite CTR as one factor, which is why I see the main component of adrank as being the eCPM achieved on your page or site. But the price they pay is very low, so they siphon off lots of valuable visitors for peanuts.
But, you might argue, doesn't the publisher still earn more because the eCPM shows the high volume/low cost of MFA ads earns more than the low volume/high cost of ordinary ads?
No, for various reasons, such as:
As the page eCPM drops, the MFA ad's eCPM starts to look better and better when compared with other ads, so the MFA climbs to the top of the list, suppressing that page's eCPM even more, and giving the MFA even more and cheaper clicks.
All in all, the presence of MFAs on your site (and in the system generally) damages your income big-time. At some point in the future I think we'll see lots of threads saying something like 'I don't see any MFAs in my sector any more, and my income is waaayy up, do you think these are connected?'
i want an algorithm that provides the highest possible payout for my sites and not the highest possible click count.
MoTi, I think your example shows how hard it is to distinguish MFAs from conventional sites.
And a further problem is that, if someone from Google were reading this, they would probably say that they already place ads according to payout (using eCPM as the measure).
But, with respect to Google, just because they build a powerful engine it doesn't mean they understand the driving characteristics of the car. The algos they design have implications way beyond their intentions. The difficulty they have is not only working out how changes they make will impact the market, but taking account of the ingenious way MFAers will respond.
There are a couple of ideas that might help - eg a minimum EPC, but for diverse sites like mine it would need to be variable by channel. I also like Harry's idea of banning ads on landing pages (excapt from SERPS I assume). But I guess it will be something that MFAers will find a way around.
But I also second the request to be able to set the minimum price for an ad. This would make the market a "real" market with "real" market forces.
Thanks
I had forgotten about the quality score and subsequent lowering in the CPC. I'm not sure if the visitor will really be affected by the listing when it comes to his/her opinion of your site. Most people know that these are advertisements and that you are not responsible for their content. If you're concerned about the perception of your visitors, as we all are, then put a disclaimer below your Adsense block that essentially states that these are outside websites and your not responsible for their content and don't approve or disapprove of them.
I also think that Google's "quality score" bot will slowly start to remove these sites as they become less and less profitable for the owners.
Really it should be mfa/a - made for adsense/adwords
As for the Content vs. Search network, interesting point! Are the bids the same for content and for search? What about AdLinks? Aren't they part of the Search network?
As for low conversion rate for MFAs, I would HOPE that Google is smart enough to realize that a subsequent click on an AdWordSense(tm);) link would count as a conversion.
Do low-converting (not low CTR) ads have to pay more to be shown?
The people with the original money is the advertiser with some widget to sell. Let's say they are going to pay $1 for a click.
Let's assume that Google pays 70% to the publisher and keeps 30%. Just an assumption, the percentages don't matter. Let's also assume that number is fixed.
So if Publisher gets and ad for a widget and Visitor clicks on that ad, Publisher gets 70 cents and Google gets 30 cents.
That's the straightforward case.
Let's now assume that the MFA gets a 50% CTR and they always go to high paying ads of Advertiser. So if MFA gets his ad on Publisher's site and gets two clicks and pays 10 cents per click, Google gets 6 cents and Publisher gets 14 cents. MFA gets 70 cents for the click and Google gets 30 cents.
So while "only" 1 dollar has come into the system the distribution has changed.
Publisher gets 14 cents, Google gets 36 cents and MFA keeps 50 cents (70 cents minus 20 cents for his ads).
That's how arbitrage works, although the numbers are totally hypothetical. Google has a huge incentive to keep MFAs, they return a huge amount of money back from white hats to Google.
EDIT: PS, I'm not saying there aren't OTHER reasons that Google should consider when deciding to get rid of MFAs, or not. I'm just saying there is a strong short term financial incentive to keep them around, AT LEAST ON THE CONTENT NETWORK.
In reality, I wonder if the MFA might have to pay for 3, 4, or 5 clicks to have a 20-33% and not a 50% CTR to get that big commission, and also not to wind up giving his traffic to another MFA. The arbitrageur's life is getting tougher, we're told. He has competition. I really feel for him. Or her.
Are the bids the same for content and for search?The advertiser has the option to make different bids for content and search. However, only content ads are smartpriced downwards, afaik.
What about AdLinks?Dunno. I pretend to be an Ostrich when it comes to AdLinks :-). They don't make me any money, so I avoid them as much as possible.
Do low-converting (not low CTR) ads have to pay more to be shown?
No, not unless something has changed since I last used Adwords (start of this year). In fact, my experience suggests the opposite. As an advertiser, my low-converting ads ended up costing me less. I dont know whether this is because it is built directly into Google's algos, or whether it is just an incidental consequence (eg: the pages where the ads tend to appear end up being smartpriced downwards because of my conversion stats) or whether it is just a coincidence. Like most outside Google, I know very little, I'm just trying to extrapolate from the data that I have available.
As for low conversion rate for MFAs, I would HOPE that Google is smart enough to realize that a subsequent click on an AdWordSense(tm);) link would count as a conversion.
What counts as a conversion is completely in the hands of the advertiser. If I had a website with a big blue button on one page saying "Don't press", I could set up Adwords so that a "conversion" was recorded everytime someone pressed that button. The Adwords advertiser defines what is a conversion.
Incidentally, thanks for reposting your workings out - it really does illustrate how the bona fide publisher is the loser in all this. Interestingly, the more clicks the MFA has to buy (from the content network) to get each conversion, the more money Google get.
PS and offtopic, RonS. If it is any consolation my Christmas Gift from Google has packed up completely - the mouse has given up the ghost.
[edited by: 21_blue at 6:01 pm (utc) on June 8, 2006]
a bigger filter list can only be a quick fix. My impression is that there must be several thousands murky URLs, so we clearly need better tools to get them under control.
One example mentioned often is the ability to set a minimum price per click. This would immediately destroy the business cases of many MFAs (they could not afford the higher prices and would simply give way to legitimate advertisers).
Another example is to block advertisers by advertiser (instead of by domain). As one advertiser may only have one Adwords account, it would be easy to remove the zillions of parked domains with nothing but ads.
But clearly, as we have seen above, Google might face other challenges, i.e. fulfilling analysts expectations.
One example mentioned often is the ability to set a minimum price per click. This would immediately destroy the business cases of many MFAs (they could not afford the higher prices and would simply give way to legitimate advertisers).
I'm not sure if I'd agree with this. In fact, that feature may actually make it easier for MFA's.
Case in point -- they set their minimum AdWords price that they will pay to $0.05 per click. Then they set their minimum AdSense ad that they will accept as $0.25 per click. Now the whole "arbitrage" activity - which is how they make a living - is done completely for them by Google. They'll never pay more than $0.05, and they'll never earn less than $0.25. The only difficult part (which probably isn't all that difficult) is finding some keywords close enough to get people in on the low clicks, make sure your content & competitive ad filter is set to only show high-paying ads on your site, and do everything you can to make sure you get a 20% click through ratio.
That's basically a license to print money.
As much as I thought I wanted this feature at first, in retrospect I think it would make the MFA problem grow exponentially overnight.
they set their minimum AdWords price that they will pay to $0.05 per click. Then they set their minimum AdSense ad that they will accept as $0.25 per click. Now the whole "arbitrage" activity - which is how they make a living - is done completely for them by Google. They'll never pay more than $0.05, and they'll never earn less than $0.25.
Well, not quite. First, this works only if the publisher does not set a minimum price. If he sets a minimum price, you would automatically kill the arbitrage. It does not make sense to buy clicks at, say, $0.20 and re-sell them at $0.25. There is always the risk that the user hits the back button.
So those publishers not setting a minimum price would further attract MFAs - those using the tool with caution, however, will see just very few MFAs. These remaining MFAs could end up in the blocking list, ideally just removing a whole bunch of URLs with one go ("block all ads from this advertiser").
Eg: suppose I were an MFAer with a CTR of 25%, I could set a max payment in Adwords of 5c, a minimum threshold in Adsense of 25c, and assuming my CTR holds at 25% it almost guarantees that I make 5c net profit on every Adsense click (I pay 20c or less for 4 clicks, from which I get 1 click that earns me 25c or more).
I didn't see mzanzig's proposal including an AdSENSE minimum pricing tool for publishers. Is there a requirement that both that and minimum bids be made available?
EDIT: Ahhhh never mind, I see how you can read what he said to mean exactly that. At first I took it to mean a minimum bid requirement for all ads from Google.
A minimum per-click value would also hurt me, where I bid low to mop up cheap ad slots to tell people about my pro-bono site.
Rgds
Damon
On the subject of minimum click price helping arbitrage, that is a point I don't think we had fully considered. Whilst it's true that this could be a scenario, there are other scenario's. I'm guessing that Google have thought them through before we have, so that's why we aren't too likely to get it.
If a lot of publishers took up the option to have psa's or no ads rather than cheap arbitrage clicks, then the side effect is that balance shifts. Cheap ads will still be in the system, but the only major place for them to be shown is on Google search. Imagine the only place for cheap ads being on their own search pages? They would soon realise just how useless arbitrage clicks are. This shifting of balance in this scenario basically puts a great big pile of poo not just in Google's back yard, but right in fromt of the TV just before Neighbours starts:) If that ever happened, they would soon sort the problem.
As I said, Google probably have thought of this scenario already and choose to try and sort it out without minimum click value.
