Re: PPC and Index tools

From: John Marshall <jmarshall_at_clicktracks.com>
Date: Tue 11 Feb 2003 18:49:22 -0600

> We use index tools to track our traffic and I have noticed a pretty
> major discrepancy in the amount of clicks that it says we receive from
> search engines and the number of clicks that the search engine reports.

Here's what I think is going on. This is based on our experience developing
a tool that tracks this kind of stuff, and then using it to track our own
advertising in Google and Overture ( cue the dog food metaphors ). We also
received some very valuable insight from an end-user.

Google and overture send the visitor through a page within their system that
triggers a click within their counter, and then redirects to your page.

The end-users browser has obviously successfully made a DNS lookup and got
the IP of Google, and managed to get pages successfully routed back.
Otherwise they wouldn't have made it this far. When they click the link, the
next request goes to the same IP as the first ( you're still inside Google )
so the chances are very high it will work. 'Click' says the counter at
Google, and sends back to the browser the actual URL of the page of your
site.

At this point, some stuff can go wrong. The new URL is probably a domain
that the browser has not seen before so it's going to need to do a DNS
lookup. That could fail. When it's got the IP, it's possible due to routing
problems somewhere on the internet that the server doesn't deliver the
page - it appears to be down to that visitor. Think how many times you click
a link to a site and it doesn't work, but a few hours later it's fine.
That's routing problems.

If you're using the page-bug data collection method there's a third thing
that can go wrong : routing problems back to the page-bug collection point.
The page did load just fine, but the requests to indextools or whoever never
make it for some % of end users.

Interstingly, there's a third problem identified by a customer of ours who
spent quite a bit of time digging into this. A browser is supposed to report
in the referrer field the last page that was not a redirect. In other words,
even though the browser is redirected through a page inside Google the
referrer stays at the original front page of Google - with all the search
terms in the URL. *However* there's a bug in MacOS versions of Internet
Explorer. The referrer in that case is the *internal* page within Google
that does the counting. For those users you don't get the nice
?q=ripe+bananas in the query string. You get a load of junk from the
internal page. This could trip up the tracking service you use. It depends
how they parse up the referring URL.

For our PPC campaigns we see about 10% under reporting compared to what
Google tells us. We live with it.


John Marshall
CEO

======================================
ClickTracks http://www.clicktracks.com

Winner ClickZ 2003 readers vote for
'Best Web Site Analytics Tool'
======================================





Received on Tue Feb 11 2003 - 18:49:22 CST