Results 1 to 6 of 6

Thread: Google AdWords statistical errors?

  1. #1
    Senior Member
    Join Date
    Jul 2004
    Posts
    548

    Google AdWords statistical errors?

    I posted here about how I think Google's advertising system may be flawed. Please read it and let me know what you think..

    And my question (apart from the ones in the post):

    Is a confirmation handshake needed for it to count as a click, or is a single request enough to trigger it?

    Thanks,

    -jk

  2. #2
    Senior Member nihil's Avatar
    Join Date
    Jul 2003
    Location
    United Kingdom: Bridlington
    Posts
    17,188
    Well,

    From the other frauds I have seen I would imagine that confirmation that there is someone there is essential. If it wasn't, AdSense would have become extinct long ago.

    There will also be control by IP range to ensure that the contact is from a plausible potential customer.

    There is monitoring for patterns, geographically, by IP address block and by date and time............they have to look reasonable as well.

    The real threat is actually from Bots, that are genuine machines and would be difficult to detect from normal traffic.


  3. #3
    Senior Member
    Join Date
    Jul 2004
    Posts
    548
    That's what I thought - there must be some way to prevent this, because if not AdWords would have ceased to exist years ago. But, I'm still trying to figure out how they control it. Even if the IP range is checked, there's nothing to say that there couldn't be a large group of people in a certain address range clicking on AdSense.. Plus, assuming the IP spoofing app was coded properly, the IPs would not be in the same range very often.

    I'd be interested to know how they control this - but whether it's bots or not, they must have some kind of solution in place..

  4. #4
    Senior Member nihil's Avatar
    Join Date
    Jul 2003
    Location
    United Kingdom: Bridlington
    Posts
    17,188
    Hmmm,

    1. Anything from outside the geographical range won't count, so heaps of crap from Nigeria are useless in the UK, and so on.

    2. People surf at different times of the day.............the number of people looking for new houses, cars and holidays between 24.00 Hrs and 06.00 hours is very small. There are known activity cycles, and your clicks had better match them!

    3. The site needs to be in the right business/interest category............like people who visit small furry animals sites don't usually click on adverts for heavy goods vehicles.

    4. A real acid test is that you have to get an affidavit from your host/ISP that you had a particular volume of traffic in a period. Now, you must surf a fair bit? How many adverts do you click on (please don't take AO as an example, as I look at a lot of adverts out of interest/curiosity/for future information)? Obviously, there are known patterns of behaviour, so if your site is beyond the normal, you won't get paid. If you had 10,000 visitors in a month only a 100 or so would click on the advert................just try claiming for 10,000.

    5. I assume that Google are simply serving a web page (the advert), and they must know if they managed to do that, similar to e-mail? So spoofing the IP address would be pointless, as the page would not be served.

    6. Google only pay you at their discretion. Any suspicion and they close your account and refuse to pay, and you have no legal recourse. You could set this up and run it for months, then end up with nothing.

    7. Do not forget that this is fraud, and a criminal offence, and that web hosts and ISPs have to co-operate with the police

    8. Sites have growth rates and activity patterns. If you fall outside the norm for those, you get bounced as well. The taxman works the same way for self-employed people, looking at averages for the area and occupation. So if you grow too quickly and show too many hits for your interest area, you are captured.

    9. Geographical anomalies also show up. I connect through Manchester or Sheffield or Leeds.........that would show up if there was a sudden peak from a particular centre, or a particular ISP.

    10. ISPs aren't fools.............you may spoof the IP address to google, but the ISP will notice the traffic, and will know exactly who you are, and it wouldn't surprise me if Google paid them for the information. That was the reason for my comment about Bots...............much harder to catch. Let's face it, you have to tell your IP who you want the "click" sent to?

    11. Bandwidth.............use an abnormally high amount, and you draw attention to yourself. Sure, your contract says "unlimited" or is a high number...........try using it regularly and you will probably get bounced by your ISP after Google and the Fraud Squad have been informed.

    12. Google actually talks to its customers. So I get charged for 500,000 clicks, but my sales haven't gone up by the normally expected margin..............oooooppppssss! It is the inverse logic of spamming, only the people have willingly solicited the advert so the % sales closure rate should be much higher.

    I hope that you get the general picture

  5. #5
    Senior Member
    Join Date
    Jul 2004
    Posts
    548
    Yup - loud and clear nihil

    Thanks. Please note that I wasn't actually attempting to do this - I was just wondering if it was possible, because then anyone might be able to do it..

  6. #6
    Senior Member nihil's Avatar
    Join Date
    Jul 2003
    Location
    United Kingdom: Bridlington
    Posts
    17,188
    Hey, J_K9 I did not for one moment think that you intended to do it. I simply thought it was academic interest or maybe you wanted some support in a debate you were having

    I worked for Price Waterhouse as an auditor, and have had experience in marketing analysis, advertising analysis and such stuff. Actually what you are looking at was going on long before the internet and e-commerce............newspaper circulations, top ten pop records?...........I was just giving you a bit of my experience and opinions.

    Sure, fraud is possible, but not the way you were thinking. I believe that you would need a Bot army (remember that you can hire them these days). You would also need some software along the lines of "cookie muncher" that would accept the advert, but kill it before it was displayed.

    You would also need some industry data or a Google insider to know what the likely trigger parameters are. You know, a bit like the old financial fraud?............all transactions over £500 have to be counter authorised so you run your scam at random values between £475 and £499.99?

    You would also need a number of identities, bank accounts and websites, as you would stick your head above the parapet that way as well

    It is certainly not an area of fraud where it is possible to make a quick killing IMO.




Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •