Google States Undetected Click Fraud Rate Less Than 0.02%3/4/2007
Jeff Howes - TrafficSentry.com
Last week, Google finally
announced it's own internal estimates of the scope of the
click fraud problem to it's AdWords advertisers. Much focus
has been directed on the 0.02% number Google defines as
"reactively detected as invalid". This number is
overshadowing the larger "less than 10%" overall number that
Google claims includes the clicks they proactively detect
and filter out before billing the advertiser.
look at the 0.02% figure first.
Closer examination of
exactly what Google said brings us to the conclusion that
the statement is cleverly worded in such a way that it means
absolutely nothing without other variables that they are not
I think that many people (including some in the press)
interpreted the statement as "0.02% of total clicks are both
fraudulent and not being detected and discounted..". What
was actually stated is that the number of billed
clicks reported to Google by
advertisers and then found (by Google) to be invalid, is
This number represents the clicks that made it through
all phases of filtering by Google, were billed to the
advertiser, reported by the advertiser, and then finally
found to be invalid by Google examiners upon investigation
after the advertiser has complained. That is not hard to
believe at all, because if Google did not detect a click as
invalid the first time, the chances are that they will not
consider it invalid upon a complaint either.
So to get to a number as low as 0.02% should be
relatively easy. First, we all know that the vast majority
of advertisers are not taking the time to file complaints,
for a number of reasons:
- The advertiser is unaware of the problem.
- The advertiser assumes Google's fraud prevention has filtered all bad clicks.
- The advertiser is unaware of the claim procedure.
- The advertiser assumes that their request will be denied, as is often true.
- The advertiser lacks the evidence to prove their claim.
- The volume/reward ratio is too low to be worth the time.
Next, out of the small number of clicks that ARE submitted by
advertisers for further review, the vast majority of clicks claimed are
rejected by Google. This is a number that they do not seem
to be disclosing and without it, you cannot complete a calculation of
their 0.02% claim.
Where is the rest of the data Google collects on our clicks?
Google likes to reject claims with the canned reply "we unable to find any conclusive evidence of invalid clicks..." (that is, if you make it past the lower level reps who's sole purpose seems to be to stall and delay you with form responses that include FAQs completely unrelated to your report.
One of the most significant deficiencies in Google's Click Quality reporting system is that they will not provide advertisers with the full click log they collect, with each individual click marked "billed" or "filtered". Why not? They do have it. It's the same data they use to make their determination. Refusal to disclose this data makes line-item
reconciliation impossible because we don't know which clicks they billed us for.
Where else would you buy multiple items over a one month period, and receive an invoice with only an aggregate total? What if you got your cell phone bill and it seemed about 30% too high but only included a total, and no breakdown of the calls made? Then, you call your phone company and they say "..Oh, we know we billed you for the right number of calls... Trust us.".
Of course, in reality, the phone company shows you the list of all
your calls and responds "Which ones did you not make?". If Google
is so sure of their assessment, there should
be no reason to hide the data from us. We give them all of our
data, and then we must blindly accept their answer without any
evidence to back it up, and generally without even any direct
response to the evidence we have provided.
Come on now,
Google. If you are going to intentionally make it impossible for
me to prove that I didn't get what
I paid for, than you need to prove to me that I did get
what I paid for.
At this point, we have already eliminated most of the clicks from
the total; those that are not reported, those that have already been
filtered out, and those which are reported but rejected (whether
correctly or not).
Finally, and perhaps the largest unknown variable, are
all of the
clicks that are both unreported AND undetected by Google.
This is the obvious flaw in Google's statement that everyone
seems be overlooking.
They are not stating anything with regards to the total
overall click fraud rate, including
both detected and undetected. Of
course they can't report fraudulent clicks they don't
know about, and I don't think anyone thinks their
filters are perfect. All of the stated numbers represent
invalid clicks that they eventually detected. Any
invalid clicks that slip through completely (and oh yes
there are many) would come out of the 90% that they
claim are "valid", and that however large the number of such clicks,
it would not affect the 0.02% figure, based on the way that it was
Perhaps this number of unaccounted for clicks makes up
the mysterious and significant gap between Google's
estimates of the scope of the invalid click problem, and
the 17-30% estimates consistently reported by
third-party click tracking services.