Welcome to WebmasterWorld Guest from 54.221.49.52

Forum Moderators: open

Message Too Old, No Replies

Fraudulent Traffic Statistics

     

speda1

12:02 am on Nov 25, 2003 (gmt 0)

10+ Year Member



I recently wrote a small script and called it from an IMG tag on a number of my pages that I advertise with various PPCSE's. The purpose was to determine the percentage of visitors referred that would load the image. Since most web users surf with images turned on, a good percentage of those that did'nt load the image would be fraudulent/bot type traffic.

I thought that others might find the results useful. So here they are. This is the percentage of good (real) traffic.

Overture 98.2
Adwords 95.6
LycosInsite 86.1
7Search 70.8
Pageseeker 70.0
goClick 57.9

Findwhat was running aroud 70% before I yanked most of my campaigns for ROI reasons.

IanKelley

5:20 am on Nov 25, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Good info Speda... and the numbers look about right.

The silly thing is that stopping the bots/cheaters in question isn't really all that hard, a lot of the smaller search engines just don't have anyone on staff who knows how to write the code.

RoboPal

10:04 am on Nov 25, 2003 (gmt 0)

10+ Year Member



speda1,

Excelent:) I wonder how wold this statistic look across different kywords and bids.

BTW if I would right a clickbot it would download images, javascripts, css etc. Let's just hope people with qualification have other things to do ;)

IanKelley,

The PPC SEs listed here do have very competent staff developers. There are several issues with the fraud protection scripts: they slow down the system, they filter off some genuine clicks etc. which negatively affects revenue. So the SEs run some filters, but allow a lot of scam clicks through.

They analize the traffic afterwords and cancell affiliates who scam them. Otherwise the number of fraud clicks would reach 100% very quickly.

IanKelley

8:04 pm on Nov 25, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Robo... I didn't post out of ignorance. I have written fraud protection scripts for numerous smaller search engines.

On a scale of millions of searches/month with 1 quality control person it is never a problem to stop the kind of fraud that Speda's images are showing (bots).

I have also seen firsthand that quite a few of the more popular PPCs do not have anything like this implemented.

You are correct that it can't all be done in real time because of speed/resource issues and the risk of filtering real traffic.

This does not mean that it's not possible to have software which makes the fraud so obvious, so quickly, that it has virtually no effect on advertisers before being removed. A lot of engines just plain don't have software like this.

I think the difference between Speda's numbers for 1st tier and 2nd tier PPCs (and then again between some 2nd tiers and others) make that obvious.

RoboPal

8:02 am on Nov 26, 2003 (gmt 0)

10+ Year Member



IanKelley,

I would desagree. It's almost imposible to filter of a good clickbot if you don't have access to logs of both the search system and advertiser. SE just redirects an http request to the advertiser's site; they can't analyze anything exept this single request. These bots don't care about the advertiser, they only want to go through the SE redirection, and that's what they do.

Bigger engines are bigger than small once :) The number of affiliates, the level of knowledge and experience of affiliates, mix of fraudulent traffic with real one in different proportions, usage of diverse and elaborate scam technologies, clickboting competitors links using SE listings on different distributors sites etc. lead to the situation when you can't always tell the difference between right an wrong. Without going into much details, I would still insist that 10% or 20% of fraudulent clicks or even more is about as good as it gets for a second tier PPC SE. They very conserned about traffic quality, but their business model doesn't let them get much better.

sb123

2:52 pm on Nov 26, 2003 (gmt 0)

10+ Year Member



Intesting data!

[edited by: engine at 6:42 pm (utc) on Nov. 28, 2003]
[edit reason] No sigs, thanks. See TOS [webmasterworld.com] [/edit]

jomaxx

5:11 pm on Nov 26, 2003 (gmt 0)

WebmasterWorld Senior Member jomaxx is a WebmasterWorld Top Contributor of All Time 10+ Year Member



Wow, those 57%-70% numbers are appalling.

A couple of points...

1. As you acknowledge, there are also occasionally valid reasons not to load images. Images turned off, an image being already cached, a page download being interrupted before it's complete, etc.

2. One problem is that the 3rd-party PPC's typically don't server up any images from their own server, and therefore have no way of knowing if a bot is downloading images or not.

IanKelley

8:18 pm on Nov 26, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Robo... I don't know how to respond to the fact that you're disagreeing with my hands on experience using hands off speculation. From where I'm sitting stopping click bots is very simple (even if logistically complex) and it would appear very simple to you if you had any back end experience with this kind of thing.

I guess I should just assume that you're arguing for arguments sake and not get drawn into it :-)

One thing though... If they're all doing the best they can, why do the numbers vary so significantly? If the task is so much harder for larger search engines, why are they so much MORE successful in stopping click bots?

speda1

8:38 pm on Nov 26, 2003 (gmt 0)

10+ Year Member



I should also note that these are stats on gross traffic received. I have no way to determine whether or not I was actually charged for the visit or whether the PPCSE was able to detect any fraud.

RoboPal

9:31 pm on Nov 26, 2003 (gmt 0)

10+ Year Member



IanKelley,

Although I don't have much experience writing fraud protection scripts for SMALLER search engines, I have reasons to believe I KNOW what i'm talking about. ;)

IanKelley

9:39 pm on Nov 26, 2003 (gmt 0)

WebmasterWorld Senior Member 10+ Year Member



Perhaps you should entertain the possibility that there are fraud protection methods that have yet to be tried by most engines (gasp) :-)

wellzy

6:16 pm on Nov 28, 2003 (gmt 0)

10+ Year Member



I agree with Ian. I have had similar to results to the ones you posted. I didn't need a script to tell me that Findwhat is worthless (just as you, I'm sure). Findwhat has to be my poorest ROI ever. Shouldn't have taken me months to figure that out though ;)

w

 

Featured Threads

Hot Threads This Week

Hot Threads This Month