| 12:30 pm on Mar 3, 2014 (gmt 0)|
What do you mean by spam rate?
| 2:20 am on Mar 4, 2014 (gmt 0)|
Adsense calls spam rate the percentage of invalid cliks.
The data exists (I know from my Adsense manager)
| 4:08 am on Mar 4, 2014 (gmt 0)|
That would be a very helpful data, as we get more and more deductions, It would be helpful which ads are getting those invalid clicks and on what ad type (text/image). Platform, and country cuts would provide even more valuable information.
| 8:10 am on Mar 4, 2014 (gmt 0)|
Very helpful indeed. It's a real shame they don't make it available in the dashboard/report interface.
If spam rate was available for adsense publishers, I bet most sites would improve (better layout, better content).
| 3:19 pm on Mar 4, 2014 (gmt 0)|
(My AdSense manager just calls it invalid clicks)
I don't think it likely. They'd have to make it available to the advertisers too, and that would no doubt scare them off.
And all these years later, they still haven't found all the ways to stomp on the black sheep, because those black sheep keep coming up with new ideas too. Black sheep are like that.
| 4:09 pm on Mar 6, 2014 (gmt 0)|
'My Adsense manager'
Just question? A bit off topic....
What do you mean with 'My Adsense manger'. You guys have your own Adsense manager from Adsense team?
It is unbelievable for my to hear it that you have your own Adsense manager.
Can someone explain it to me.
| 4:35 pm on Mar 6, 2014 (gmt 0)|
When you get big enough you start to learn that Google isn't so unreachable after all and things like "My AdSense Manger" become available to you.
|wa desert rat|
| 6:15 pm on Mar 6, 2014 (gmt 0)|
Adsense has gone through several iterations of PPC fraud over the years. The publisher-based fraud where a web site owner would get his pals to click on ads was the easiest to get rid of. Then the cross-clicking groups of publishers were pretty easy to track down and get rid of.
Now we have the middle-men PPC fraudsters using sophisticated command and control bots installed on exploited Windows machines that, like the distributed-denial-of-service botnets (DDoS) are extremely difficult to detect and shut down.
When these bots targeted only ads that profited the middle-men they were easy to spot. So now the C&C servers give them a lists of ads that also contain random ads that don't profit the middle-men.
When the bots showed up on a site and immediately clicked and left that behavior was trackable. So now the bots show up and "read" a bunch of pages before clicking.
The java code from Google that we place in our html gives Google a lot of insight into what users do on a site. There are some analysis tools that we can install that mimic this and it's pretty interesting to drill down into visitor actions. I now have one that gives me a "profile" of a user over time; how many times they've appeared, how many pages they read, what pages they read, how long they were on... all over time.
Here is an example of part of a user profile:
Spent a total of 12 hours 59 min on the website, and viewed 250 pages in 9 visits.
Converted 0 Goals.
Each page took on average 0.11s to load for this visitor.
5 Mar 2014 - 1 days ago
from: Direct Entry
6 Mar 2014 - 0 days ago
from: Direct Entry
9 visits from Riverside, United States United States (show map)
This also includes a complete listing of what pages were viewed on each visit as well as the IP address. So I can see everything Google sees... except who clicked what and when.
If I had that information then I would be in a better position to detect spammers (and ban them) than if G just gave me spammer lists. And, anyway, given the distributed nature of a botnet, and the fact that the PCs are often on DHCP IP addresses (which can change), I'm not sure any spammer list would be accurate for long.
| 6:19 pm on Mar 6, 2014 (gmt 0)|
|That would be a very helpful data, as we get more and more deductions, It would be helpful which ads are getting those invalid clicks and on what ad type (text/image). Platform, and country cuts would provide even more valuable information. |
It is possible to ascertain this information but it takes some stat-watching. I'm not saying this is practical or better than having G do the work for you but here goes.
Open a performance report for today. If you're looking for ad units that may be causing problems use the Ad Unit report. Now start recording the numbers for each ad unit in 15 minute invervals (suggested) and watch for which ads are loosing clicks. It's that simple. Not perfect but it can be effective. You can do the same thing with a country report.
I don't see a lot of takebacks so it's not worth my time but if I was, I'd be looking for which ad slots were most vulnerable, what types of ads are getting hit (image vs display). wa desert rat is already reporting a difference in this activity between mobile and desktop. I'm betting if people really start looking at this, we're gonna discover some very predictable patterns. Top banner ads would seem to me to be the highest priority target for a number of reasons but it's just a guess on my part. We may find that image ads are more vulnerable than text ads. Who knows? Only people observing data can make these observations. The more the focus of these discussions moves from the bottom line to the details that make up the bottom line, the more we'll learn about this.
I'll ad an observation of mine that may or may not factor into this discussion but is worth mentioning. I build my sites from scratch in ASP.NET. In other words, I'm not using an out-of-the-box content management system. This may actually have something to do with why I'm not seeing this problem as much as others. Every Word Press site has something in common with every other WP site. Maybe it's easier for these bots to attack systems they have been taught to understand (structurally). It's easier to program a bot to look for a sidebar widget for example that to try and learn some custom page structure.
Also, because my pages are dynamic, I use different size ads and locations on what is essentially the same page (maybe just showing a different category for example). These are basically built from the same script but ads move around the page depending on the length of the content in each column. Predictability of your page layouts could be making it easier for bots to keep hitting the ads on your pages. Like I said, this may or may not have any bearing on the issue but it is something I do differently than your average WP publisher.
| 10:15 pm on Mar 6, 2014 (gmt 0)|
I see about 50% of my adsense revenue be removed everyday, I guess google thinks those where invalidate clicks.
| 9:28 am on Mar 7, 2014 (gmt 0)|
|There are some analysis tools that we can install that mimic this and it's pretty interesting to drill down into visitor actions |
NSA - eat your heart out!