Forum Moderators: martinibuster
In the last couple of months I've spent less time looking at the ads on my site, and have not blocked that many ads. I know Google are working on algorithms all the time, and some work has been done on quality scores in adwords. Maybe it's time to conduct an experiment in removing the list to see how/if the algo's have improved. I've removed all except ebay - that IS a step too far :)
I am also using a tracker to see what ads are clicked. That works OK except it doesn't tell me what ads were clicked from an adlinks unit.
It's too early to tell anything from the overnight results. However, I can report back on the ads I'm seeing. Firstly, I'm seeing 3-4 ads per unit. That may not be significant, as over the last few weeks I've seen mostly full ad units.
The other point to note is that the target bot clearly hasn't been improved. I'm seeing an MFA in the top spot both last night and this morning. Same viewing the US via a proxy.
The tracker shows that nobody has clicked the MFA - so far there have only been clicks on genuine ads.
It's too early to say if there are any financial implications, but I have to say that with MFA's appearing this is inevitable.
My gut feeling is that blocking is the correct thing to do, as even if nobody clicks the MFA's they are taking up space of genuine ads that people might click. I'll let the experiment run for a few days and see what happens. Obviously I'll report back here.
I think 500 would be a reasonable number. How about it ASA?
Can you swing that for us?
And my figs are very stable to around 3 to 7 percent normally in variation.
So I suspect the real figures you are looking at are actually worse than you think!
Since the beginning of this month and the adsense down/targeting or whatever my stats were down by 15 percent. As from today its all caught up I think, as I expected it would frankly judjung by two years figures!
David, would you PM where I can get the tools for AS click tracker? I am looking for a similar tool for further analysis my AS clicks. Thank you.
As a result, one of them is now being shown on more than 40% of pages on my site, but I appear to be making a little bit more money. My ECPM was up for the first two days of the test but today - a Sunday - it is back where it used to be.
According to the tracker, MFA clicks are down to 10% and one of the ads shown (and clicked) I would include in this is actually just bad targetting. So MFA's are taking up a lot of space and getting a small portion of clicks - wasting space that could be better used by real ads.
Overall site statistics are ctr is up by 41%, and epc is down by 8%. I believe that these figures are skewed by a blip of good fortune on an adlinks unit on one of the pages that usually is ok, but not spectacular. None of the other adlinks units are up, so I can't conclude that dumping the blocklist has caused this increase in the one unit. I think somebody has posted a link to the page on some forum or other causing an increase in traffic on that page.
My main ad unit is probably the best indicator, as that's where the biggest proportion of clicks and income historically comes from. Clicks are down 30% for the time of this experiment and income is down by 40% over the same time. This is consistent with the fact that real ads aren't being displayed, the percentage of clicks on MFA's is lower and no doubt the payout is too.
I'm going to let it run until tonight to gather some more statistics. Although it would be nice to let it run for a week to be more certain, I can see how it's going and I can't afford to take a long term hit for the sake of an experiment.
I think the main thing I've learned by this is that blocking MFA's does indeed increase earnings, and that the block list doesn't have to be that large to be effective.
The key thing that you suggest that gives me heart is that the block list does not need to be large to be effective.
In other words, you should be able to stop playing whack-a-mole LONG before you fill up your ad filter.
Just ban the most egregious and the 80/20 rule suggests that you will have gained most of the benefit.
But G really would help themselves if they showed which entries in the filter we no longer effective for one reason or another, and/or would allow a ban by publisher ID also to kill multiple sites with one entry.
Rgds
Damon
I'm not encouraging you to carry the experiment forward- we all have to make a living- but I'm just throwing it out there that a weekend test may not be reliable for drawing a conclusion.
Me!>>>
Well that depends on your site! I have 15 seperate sites and subjects. Thats a potential 15x as many MFAs etc.
I have onluy blocked persistent ones, on 4 out of my 15 sites, the biggest earning ones. And now the filter is full. That is US and UK only. There are yet masses more of the things on the canadian region choice! These I cannot get rid of, and the same with the rest of my smaller sites.
If everyone blocks as many as possible though then they will soon die out because it will not be profitable to scavenge from us in this way.
I expected your result though as it is exactly what I see when I try the same thing.
I will do the same in a week or two (now is a bad time as my earnings became strangely unstable at the beginning of this month and are only now beginning to come back to normal levels. Once they are steady again I will dump the lot from the filter and wait until the result is stable. I already know what to expect though because I tried this before several times!
Your test didnt allow for
a)lower conversion and subsequent smart price cpc drop.
b) your visitors buggering off because of the "crappy links" and so on!
So the real long term results would be worse than you already found I think!
I take the point that a weekend test is not necessarily going to prove a lot, and I would be wary about forming any deffinite conclusions from the statistical data. Just for the record, ctr is up by about 20% over a normal weekend, epc is down by about the same and overall earnings are down by about 20% on a normal weekend. But as I said before, I would be cautious about reading too much into the statistical portion of the experiment.
From the top post of the thread:-
In the last couple of months I've spent less time looking at the ads on my site, and have not blocked that many ads. I know Google are working on algorithms all the time, and some work has been done on quality scores in adwords. Maybe it's time to conduct an experiment in removing the list to see how/if the algo's have improved. I've removed all except ebay - that IS a step too far.
I started the experiment principally to see if the quality scores and any targeting algorithm changes introduced have helped in filtering MFA's. The simple answer to this was evident within a couple of hours of dumping the list. Google's algorithms have not improved. MFA's were back - big time. Many of the top spots were taken by MFA's, with advertisers I know work well being dumped. I know their ads were historically clicked on thanks to the tracker.
I also set out to see what ads were clicked with the benefit of a tracker. That data is really inconclusive, as this weekend a lot of my traffic was (unusually) due to adwords, and the tracker doesn't indicate what ads were clicked on adwords pages.
Nitrous:-
Your test didnt allow for
a)lower conversion and subsequent smart price cpc drop.
Unless the results of the weekend were unbeleivably good, I had no intention of letting the experiment run for long enough to be penalised by smart pricing. I know only too well what that effect is.
I started blocking last July 4 after watching earnings slip by half over the space of 2 months. The slip co-incided with two MFA's appearing. Removing the MFA's started a positive cycle of higher earnings, and higher EPC reward by smart pricing. The main MFA's I saw in this experiment were the same two, but there were some others I have blocked appearing a lot as well.
b) your visitors buggering off because of the "crappy links" and so on!
Again, I know what the effect of these are, and the experiment could not be allowed to jeapordise that. The tracker shows me that visitors who click the ads are often the ones who have the highest number of page views per visit, and frequently were return visitors. I didn't let the experiment run long enough to asses the result on return visits. I don't think I need to to know!
My conclusions from this experiment are mainly that Adsense technology has not improved. Quality score initiatives and targeting have not changed to prevent MFA's showing.
My other conclusion is that my original decision last July to block MFA's was absolutely correct. The limited amount of data I have from this experiment mirrors the circumstances preceding the original blocking. High ctr, low epc and earnings. I accept that the data is not enough to draw firm conclusions due to the short duration of the test.
It's been an interesting experiment, and that's how I'd describe the results - "Interesting" rather than conclusive. However, adsense replaces my wife's job so I can't afford to continue it!
To risk income with an experiment is something very unusual.
as i pointed out earlier, that's not how it works.
ebay was mentioned in this thread, but ebay is not an mfa... what makes sites like ebay, shop dot com, etc., undesireable is that they use any keyword they want in their adwords ads.
a lot of mfa's work the same way, in that they can use any keyword they want in their ads, so they can show up anywhere they want... google can only match adwords keywords with publisher web page keywords, how could there be anything beyond that?
i have seen mfa's that don't have any matching keywords in their ads, which is something that google could easily stop if it wanted to... it's the simplest aspect of contextual targeting you could ever ask for.
so google is deliberately letting mfa's with no matching keywords show up as a small percentage of the ads, because they make money off of it... good for google and good for the publisher is not always the same thing.
which is why google refuses to increase the competitive filter limit, won't allow us to see what ads are clicked on, etc.
[adwords.blogspot.com...]
Why are we doing this? Simply stated, we always aim to improve our users' experience so that these users (your potential customers) will continue to trust and value AdWords ads. Have you ever searched on a keyword, found an ad that seemed to be exactly what you wanted, and then clicked on it only to find a site that had little to do with what you were searching for? It's not a great experience.Incorporating landing page assessment into the Quality Score will help us improve the overall advertising experience for users, advertisers and partners by increasing the quality of the sites we present in our ad results.
Advertisers who are providing robust and relevant content will see little change. However, for those who are providing a less positive user experience, the Quality Score may decrease and in turn increase the minimum bid required for the keyword to run. To help define site quality, we've created a general set of website design tips and guidelines that should help you evaluate and optimize your site.
This is part of what I was trying to test - to see if the above would stop MFA's that have no content whatsoever (just adsense blocks) appearing. My reading of the quality score initiative above is that it's supposed to give priority over proper content landing pages, or at least put up the cost of advertising to the advertisers that have poor landing pages as described above. I can categorically say that it's made no difference whatsoever. The only way I can keep a site that has no content other than adlinks / adwords blocks off of the top spot of my website is to block it.
Come on Google - explain this.
If the ad quality score worked, then irrelevant Ebay affilliates (new and used dead pope's etc) should not appear on anyone's site - but they do.
I really don't see why Google think it's a good idea to ignore the feedback we give them. In the months I've been blocking, I've made them a lot more money on my account than they would have otherwise had.
The experiment is running on longer than I hoped. The block list has not yet kicked in, and I'm still seeing MFA's and the epc is now at half of normal for a weekday.
At this moment, I'm seeing three ads in my main block - two of which are MFA's. The top spot is the one that has a single ad in the middle of a page of white, and some keywords way below the fold. Why is that top Google? The second ad is one of my regulars, and the third is a new MFA that I've just blocked. If the quality score had worked it would not be shown - not only is it an MFA, it's also completely irrelevant.
I really don't see why Google think it's a good idea to ignore the feedback we give them. In the months I've been blocking, I've made them a lot more money on my account than they would have otherwise had.
Google's assumption is that you are the exception. Most publishers would not have the time, initiative, and wit to do what you have done. They would rather get a system-wide approach to maximizing income working than let publishers fine-tune and possibly screw up and make things worse, or figure out a way to game the system.
It's like the site-targeting--they put ads on a site that has been site-targeted if their algorithm predicts that the site will make more money from the impressions then they would have made from clicks. But that prediction is based on averages, not history of individual sites. So for some sites, it won't be accurate, because they are doing much better than average.
Google's assumption is that you are the exception. Most publishers would not have the time, initiative, and wit to do what you have done. They would rather get a system-wide approach to maximizing income working than let publishers fine-tune and possibly screw up and make things worse, or figure out a way to game the system.
No exception, we all know the the new and used ads for imposibilities. While I can't click on my own site, I make a point to click all of them (whithout clicking further) on any other site I visit. My way contrubution to the real site and my way of saying what I think of the scammer.
What google really has to make available to us is a filter based on e.g. a regexp where we can ban any add that uses words like amazon, ebay, new and used, instead of having to hunt down the scammers in all possible geotargets.
Moreover the tool they provide is MSIE only, not so great if you know many webmasters use a mac.
The simplest trick to stop them once and for all: you get an incoming click on adwords, fine, amount noted. If you get an outgoing click you will at best earn $0.01 less than the incoming click was worth to you.
End of story with the scam.
Buying traffic to sell it back to the one you bought it from . No matter how you turn it there is no added value, yet they are allowed to make a profit. If it walks like a scam, if it talks like a scam, ...
No exception, we all know the the new and used ads for imposibilities.
Who do you mean by "all"? If by "all" you mean the few hundred WebmasterWorld AdSensers who read the posts on this forum regularly and carefully, then I agree. But "we" are only a small fraction of AdSense publishers. Don't over-generalize from your personal experience....
Who are the Google people that do this "assessment?"
There's been a rash of bogus "just complete our survey" ads promising all kinds of "free gifts" hitting my sites (as well as many other publishers). There's no legit landing page here - it's just a scam to get email addresses. If bogus advertisers like this WERE actually getting assessed, then their ads wouldn't/shouldn't be everywhere, right? But they are.
[edited by: Play_Bach at 4:48 pm (utc) on Mar. 14, 2006]
While I can't click on my own site, I make a point to click all of them (whithout clicking further) on any other site I visit. My way contrubution to the real site and my way of saying what I think of the scammer.
I may have gotten hold of the wrong end of the stick here, but doesn't doing this contribute to David-UK's point that Google displays these types of ad more often because they "do well" on other sites? So if you and everyone else clicks on them, why shouldn't Google assume these ads are "better" and display them wherever it can?
Surely the best way to beat them would be if no-one clicked them at all...
that amount of mfa garbage is disgusting, and i don't understand why it would happen with your top ad block.
1) do you get a lot of mfa site targeting
2) if so, is your ad(not page) cpm close to the typical $1/per thou banner cpm... i'm not asking for specific numbers, but there has to be a reason why mfa's are showing up so much, especially in the top block.
thanks for taking the time to point out the so-called "quality score"... i'm trying to imagine what standards google could possibly use to base such a system on... the mfa'ers appear to be much smarter than google about that :-)
one thing to remember here is that when you have pushed hard enuf on adsense support about this mfa trash, their last line of defense is to say that they will forward your complaint over to the adwords people... it's a tacit admission that there is nothing they can do to stop the garbage.
if you haven't gotten that far with 'em, then you haven't pushed hard enuf... get back on the keyboard, and let 'em know that this mfa garbage is unacceptable to honest publishers who care about the reputation of their sites.
It would be interesting if we could post and share our block lists somewhere to see if there was a common theme.
1. Ad targetting has been good since restoring the list. The adlinks performance is increased too, so I would advise people to look at what ads show on adlinks pages as well as in main banners. I'm less bothered about what ads show on adlinks pages as the visitor has actively selected to see adverts, and to an extent will allow some of the ads I consider to be dubious MFA's. The real baddies still get zapped though.
2. Yesterday was my best day for a long time. Traffic and clicks normal for a weekday, but fractionally down on Monday and Tuesday. CTR average. EPC highest it's been for a while, ecpm and income up. I don't want to attach anything to this though - it maybe just a statistical blip.
3. I emailed Google via the adsense website contact form informing them of this thread, and giving them details of the worst offending MFA's that had replaced regular adverts that were displaced. One of the ads was that of a personal freind of mine who could not understand why I had dumped his ad for a junk ad. I explained the experiment to him, and he thought I was just plain mad. Thankfully his ad has returned since restoring the blocklist.
Google's reply after two days? THEY HAVE NOT REPLIED. I did ask them to read the thread and to send back a considered response to my questions as opposed to a canned response, but no reply.
I might try emailing them again later in the week, but having researched and typed out all the information once, I'm not that inclined to repeat it.
I made the case for Google to recognise that blocking MFA's increases Google's and our income, and also increases the credibility of adsense, which has to be to everybody's advantage except the MFA's.
I also made suggestions for tools to deal with blocking MFA's and effectively managing the block list.
I asked specific questions on the target bot and quality score.
I don't think it's unreasonable to expect some form of resonse - even if it is the canned "Our technology is never wrong - it's you that is the problem" email from the Google fortune cookie generator adsense support use to reply to emails.
I don't think it's unreasonable to expect some form of response - even if it is the canned
Google's strongest principle is it's stance against invalid clicks, a stance I (for one) can understand. I agree fully that the long term success of the program does rely on advertiser confidence and that the only way to maintain credibility is to deter and also eliminate situations that lead to eliminate invalid clicking.
Likewise you would reasonably assume that Google seeks to maintain credibility with the audience that is being targeted by eliminating unreliable ads. Instilling confidence into the audience targeted by the ads and taking a pro-active approach in preventing ad blindness should be a high priority objective also.
Clicks are down 30% for the time of this experiment and income is down by 40% over the same time.
if i understand right, for your main ad block, your epc AND ctr have gone down while deleting the ban list? could you please make your stats more clear, since this is a very crucial and interesting experiment.
if so, you are a lucky guy. the problem on my sites is, if i put ebay, shopping, mfa et al into the ban list, my ctr drops significantly by about 30-40%. increase in epc can hardly cope with that. demographic of my visitors is, that they actually do excessively click on mfas, price comparison sites with even more ads, irrelevant crap ads with catchy ad copy, scam ads etc.
their landing-page experience is for sure desillusioning, but the click took place and the money has been made. this is the crux! ad quality, network reputation (as claimed) and long time effects definitely aren't considered by the adsense algo.
publishers AND visitors need to be protected against these ads! since this is obviously not googles' objective at this time, they should rename the competitive ad filter to "crap ad filter". that would be at least a kind of admission to failure.
Clicks are down 30% for the time of this experiment and income is down by 40% over the same time.if i understand right, for your main ad block, your epc AND ctr have gone down while deleting the ban list? could you please make your stats more clear, since this is a very crucial and interesting experiment.
Sorry for not making this clear. Yes ctr was decreased as well as epc. However, this was different to the time before I started blocking. Then ctr was very high, but epc was right down the toilet.
I would say that this experiment was not so much about the metrics, as the time allowed for the experiment was short. It was a test of Google's targeting and quality score algorithms mainly. The long term implications of dumping real advertisers and replacing them with what is essentially spam is not in dispute - it's a matter of fact.
publishers AND visitors need to be protected against these ads! since this is obviously not googles' objective at this time, they should rename the competitive ad filter to "crap ad filter". that would be at least a kind of admission to failure.
IMHO, Google should accept the fact that MFA's exist. They should have a competitive ad filter as they do now, but have a separate spam bucket with unlimited number of entries. They could then crawl people's spam buckets and use that information to identify and investigate who we as webmasters classified as spam. Thet wouldn't necessarily mean booting people from adsense, but simply a trigger for investigation.
The competitive ad filter has a legitimate purpose, and this should be kept separate from ads that are essentially spam.
Firstly they did read the thread, and they do take the discussions here, and in other forums seriously. Usual assurances about their technology etc.
The interesting point is that whilst there were no comments about MFA's (and nor would I seriously expect there to be), they did say that in the future they may be able to expand filtering options. Now what is meant by this is anybody's guess, but it does suggest they actually are aware there is an issue, have acknowledged it exists and are working on solutions.
We await developments! I note the specific MFA's I reported in the email are still online and serving ads.....
Allow publishers to set the minimum profit per click on their sites. It's the easiest way to block MFA parasites.
Yet I still get some low-value clicks. I really don't think they are to MFAs. I believe they are to real advertisers who are simply in a low EPC area--I get ads from a number of different areas on my site.
Based on this experience, I think if I set a minimum EPC I would block legit. advertisers and lose some income....
Besides, I'm not in the business of assisting low bid advertisers, no matter how legit they may be.
