≡ Menu

Somebody Get Me a Statistician, Please.

As traffic has been increasing on this blog, I’ve been looking for ways to monetize it effectively.  One of the biggest hassles has been that huge discrepancies exist between the various website tracking engines out there, and much of the information is just plain wrong. For advertising networks that depend on these engines, and for individual site owners like myself, this is a real problem, since pricing is driven from these statistics.

Take a walk with me through my statistical garden…

On Technorati, Saunderslog.com is listed as having 1,069 links from 379 sites. Google says 710 links.  By comparison, Alexa shows just 32 links.  Alexa also claims I own a bunch of porn sites in addition to Saunderslog as well.  I do own one other site, and it’s not a porn site… it’s a suprise for my wife for our 20th anniversary. 

These numbers are important because they drive search engine ranking.  I’m not sure whether the Technorati number is correct, or whether the Google number is correct, or how they’re calculated respectively.  Alexa is clearly wrong.

Webalizer, the statistics engine that comes with my blog hosting, shows about 6500 visitors yesterday.  Awstats, which weeds out bots, feeds, 404’s and so on, claims 2410, while Google Analytics calls it at 1542 visitors, and sitemeter says 1500.  Who’s right?

And what about the advertising numbers?  This site has three kinds of pages: multi-post pages which aggregate a bunch of posts (like the front page), story pages (long pieces that are designed to persist), and single posts.   

Google Adsense, yesterday, showed these statistics:

 

Total Impressions
2,379
Multipost Bottom of Post
289
Multipost Sidebar
394
Single Post Blob
1,979
Single Post Bottom of Post
1,959
Single Post Footer
1,967
Single Post Header
2,014
Story Blob
7
Story Footer
7

Clearly, the multi-post pages of the site (including the front page), get very few views. Perhaps that’s because it’s ugly, as Om Malik once told me over drinks…  However, if we focus on the single post pages, each of those runs exactly 4 Google advertisements.  A text link bar at the top (labelled Single Post Header), a blob at the start of the story (labelled Single Post Blob), a banner above the comments (labelled Single Post Bottom of Post), and a banner below the comments (labelled Single Post Footer).

Notice that the header block claims 2014 impressions, while the footer claims 1967.  All the ads are loaded whenever that page loads, so why the discrepancy?  I buy the argument that readers might not ever scroll to the ad blocks at the botto of the page, but how can adsense determine this?   Moreover, how is that Google Analytics can claim 2635 page views, while Adsense claims 2379 total impressions.  Does that mean Google is not loading any advertising for 256 of those page views?

Confused yet?

And, just to add a little more confusion to the mix, Adbrite, which I am experimenting with for advertising, runs on my multipost pages and claims 510 page views per day, from 200 visitors. 

Here’s how I’m thinking about it.  

I don’t believe Alexa’s results.  Period. I don’t own any porn sites, and there are a bunch more than 32 links coming into this site. Alexa’s traffic numbers (expressed in reach per million, and so on) are hard to compare to anyone else’s because they require me to know how many millions of users there are on the internet.  For links, I’ll stick with Technorati, and I’m not going to focus on Alexa’s reach numbers, at all.

Since I can’t be sure whether the advertising is being loaded or not, I am simply going to ignore Google Adsense’s numbers.

Webalizer clearly overstates the number of visitors I get, since it includes bots and so on.  There’s a wide gulf, however, between Google Analytics and Awstats.  Sitemeter and Google Analytics corroborate each other, which is unsurprising since both require code to be inserted in your page. However, neither Google Analytics, nor Sitemeter, include the readers coming in via the RSS feed.  Feedburner shows about 650 feed readers on my main feed any day, but doesn’t record statistics for the secondary feeds (I have to fix that).  If you add the 1563 viewers claimed by Google yesterday and the 655 readers claimed by Feedburner, you get 2218 viewers, which is a difference of 192 from the 2410 claimed by Awstats. 

For me, Awstats wins the day.

And me? I’ll be redesigning the site to provide consistent advertising space across the whole site.  I’ll experiment a little longer with Adbrite and Text-Link-Ads but if they don’t produce results (likely because they sell the space using the wildly inaccurate Alexa statistics), I’ll drop them. 

{ 4 comments… add one }

  • skibare July 15, 2006, 8:04 am

    I have NO idea why MORE bloggers do NOT resell VOIP to the consumers like I do to ""MONETIZE"" the Blog!
    its FREE and it WORKS…………the MONEY comes IN, I do nothing………
    skibare

  • WhoopJack July 15, 2006, 10:53 pm

    I'll take a stab in the dark at explaining some of the differences in numbers between the differently placed google ads like the single post header and footer. Google cannot always find an ad to match the page quickly enough and in those cases it returns what they call a Public Service Ad, unless you specify an alternate ad by adding something like 'google_alternate_ad_url="http://blah.com/alt_ad_1.html"' to the adsense script block. When they display these PSAs they do not count the hit. I see this in some of my pages where I have a text based ad at the top and a graphical ad at the side. The graphical ads often don't get matches based on the content of the page and therefore are not returned nearly as much as the text based ones. Thus the graphical ad hit counts are usually lower than the text based.

    As a thought on the differences between the google analytics and adsense totals, possibly this could be due to more ad blocking software being configured to block actual google ads as opposed to the analytics script block.

  • Jason Drohn July 16, 2006, 10:29 am

    A congratulations is in order, primarily. Your blog is very impressive and the visitor stats are amazing. I am new to the blogosphere (prompted by a few posts on how to start a blog- namely Randy Morin's and your own), but enjoy it immensely.

    I don't have much advice on the blogging side of it, but i have been into domaining for about a year and a half. Having used 5 different ad aggregator services and seeing mixed unique hit results (and drastically different profit percentages), I think it is safe to say that none of them is perfect, unfortunately. There are so many tracking statistics and algorithms that it is immensely difficult to narrow it down to one of two. Each site or company has its own protocol that they use.

    I have seen some advertising for beta versions of stat companies that actually record user's interaction with your site, and you may view them in a movie format. I think that that sort of tool will be hugely beneficial once the bugs get ran through it.

    It would be wonderful to think that there might eventually be a type of standard practice, an IEEE type of protocol, but unless you and I start it, I think it is wishful thinking! But I am up to entertaining that notion!

  • Alec July 16, 2006, 6:29 pm

    Skibare: thanks for the tip. You have a new downline.

    Whoopjack: makes sense. Generally, I think the whole thing is way too complicated.

    Jason: I too am waiting to see what these more advanced tracking sites will do.

Leave a Comment