Starting June 19th, we're allowed to submit claim forms to Google Adwords for click fraud expesnses charged to us since Jan 1, 2002.
they've created a $90M fund to pay us back for faulty charges.
I was confused with one part of the form, however. They ask you what % of your total clicks do you believe to be fraudulent.
How are we supposed to determine this?
Any suggestions?
thx.
I was confused with one part of the form, however. They ask you what % of your total clicks do you believe to be fraudulent.How are we supposed to determine this?
If you have access to the web server logs, or some other records that contain the traffic sent to your web site, you can submit them to Google. Look for things like repeated IP addresses (or address blocks) in very small time frames, sudden traffic spikes, clicks on the landing page only vs. exploration of the rest of your site, and other things you consider anomalous.
Nobody. Small companies rarely keep those. For large companies, log files can be a GB per day (or more). They rarely keep these past a few days.
Which raises the question: what will Lane's lawyers get? $30m (their 1/3rd of the max possible payout) or 33% of what actually gets paid out? If it ends up that only a few million dollars are paid out, the lawyers may end up with nothing (after covering their expenses).
Do the laywers get their payment in half credits towards future advertising? That would only be fair. Also, hilarious.
The page for registering is at www.clicksettlement.com
The claims page only asks for name, address, acct #, amt of adspend during the period, etc. It also asks you to estimate the amt of clickfraud. You pick a range from 1% to 100%. It doesn't ask how you estimate or prove that.
Accuracy and clarity are extremely important. This Claim Form is a filing with the Court and any false filings may lead to serious penalties.
Wonder if the court comes after you if you have to guestimate how much bogus traffic you got.
It says estimate. All estimates are, by definition, incorrect.
The minimum percentage of clicks which can have been fraudulent is by definition zero, unless you have some really strong evidence otherwise.
The maximum percentage fo clicks which can have been fradulent is equal to 100 less your actual conversion rate, e.g. 100% - 5% = 95% for a site making 5% conversions from adclicks to sales / signups etc.
It would be a reasonable estimation, as no other information has been provided by Google, to take the mean of these two figures, e.g. (0%+95%)/2 = 47.5% estimated click fraud.
When forming a single value estimate it is accepted practice to find the upper and lower bounds of possibility and choose a value centrally located to that range. In that way you are showing no bias in either direction.
Worked example 1:
Jim's widget shop sells widgets online. During the settlement period he purchased 1000 clicks. His tracking indicates that those clicks resulted in 400 sales.
The evidence indicates that at least 400 of the 1000 clicks were genuine. There is no additional evidence that the other 600 clicks were or were not fraudulent.
The number of invalid clicks lies between 0 and 600. An appropriate unbiased figure to choose is 300. This gives an overall percentage of 100 x 300/1000 = 30% fraudulent clicks.
Worked example 2:
Fred's widget news site uses Adwords to generate signups for his e-widget-newsletter. During the settlement period he purchased 500 clicks. His tracking indicates that those clicks resulted in no new signups. His manual analysis of the logs shows repeated clicking from three IP addresses accounting for 250 clicks.
There is no evidence that any of the clicks were genuine. There is evidence that 250 clicks were fraudulent.
Hence the number of fraudlent clicks lies between 250 and 500. An appropriate unbiased figure to estimate is (250+500)/2 = 375 invalid clicks.
This is an overall invalid click percentage of 100 x 375/500 = 75%
