Search Longtail - Hitwise Data

How long is the long tail? Bill Tancer of Hitwise wrote an interesting post on the Hitwise blog about the very subject. Using 3 months of their data, he pulled more than 14 million search terms. Though they are looking at a small and limited data set, there are some interesting findings:

• Top 100 terms: 5.7% of the all search traffic
• Top 500 terms: 8.9% of the all search traffic
• Top 1,000 terms: 10.6% of the all search traffic
• Top 10,000 terms: 18.5% of the all search traffic

What I found surprising, this means that the top 0.07% of the terms generate nearly 20% of all search. Tancer says, "In summary, the long tail aspect of the search is true, but the data tells us that there may really be no head or body." which I agree with, but I think he misses the concentration at the front. The Long Tail is supposed to be a steady curve to 80/20. To say there is no head is not quite accurate. Flattening out to a longer tail than was originally postulated would be correct but you can't dismiss 20% of traffic in the first 1/10 % of search terms.

I think it's also noteworthy to mention that the slice was 10 million US specific users and no adult terms. (why does everyone try to ignore the adult traffic, as if it doesnt exist?)

Link Profile Analysis

This is really a great post by Aaron on ways and tools to look at your site's link profile. There are also some nice historical anecdotes in there. I was surprised to find 2 gems that I had somehow never seen before:

This link analysis tool

And this independent web crawl and free index

Google Testing - Forum Search Results

Google is testing this in the SERPS for any result which they think is a forum:

Click - New Book by Bill Tancer

Bill Tancer is general manager of global research at Hitwise. He just wrote a book called Click, What Millions of People Do Online and Why It Matters. He was featured on Good Morning America and 20/20 yesterday, Aug 28.

The book sounds interesting and it's written by a data analyst which is a plus. I found this humorous, you can read the first chapter, (PPC) Porn, Pills, and Casinos here. He talks about the prevalence of Porn in daily users habits, does a comparison of the search histories of 'online poker' vs. 'sportsbook' and the blue pill.

Global Page Load Time and Browsing Habits

I looked for this data a couple months ago and it was not easy to find. ClickTale, the hosted, live analytics company compiled usage data of their subscribers to come up with a pretty good overview.

Both posts are worth a read, but here's a summary:

Part 1:

• There is a good graph of main country average pageload times
• Graph of on page time per country
• Countries with slower load times, users spend disproportionately MORE time on each page
• China has the lowest speeds
  
• Dutch and Israeli surf the fastest

Part 2:

• Pageviews per country
• Overall average time spent on 1 site per country

Web Habits Across the Globe Part 1 and Part 2.

Googler Comments on Links at Google Groups

JohnMu added this comment to a Google Groups discussion about a site owner who thinks his t-shirt website has been penalized. It's looks like the site is indeed suffering from a penalty of some sort and does not rank well for 'Web Site' or 'WebSite' or 'WebSite Product', but does return a #1 result for 'WebSite.com'.

JohnMu's response is enlightening, especially reading though a bunch of rambling theories from various commentors.

It might be that the links to your site are not counting the way they
might have in the past. In general, it is important to us that links
are not just exchanged, bought/sold or otherwise used in an attempt to
manipulate rankings, as we have detailed in our help center article at
http://www.google.com/support/webmasters/bin/answer.py?answer=66356

If you find that your site has issues with regards to our Webmaster
Guidelines that can be resolved, I would recommend doing that and then
submitting a reconsideration request, detailing the changes that you
have made.

These guys seem to have a failry large % of links with similar link text, in sidebars, sitewides, and on blogs. It also seems that they have a large amount of paid posts with very targeted link text.

'not counting the way they might have in the past' is a general way of phrasing that I have seen Google use before as a response. I think most have taken this to mean that the links have been 'devalued' in some way and maybe the act of devaluing really is a penalty. This must be algoritmic in other words there is a threshhold that is passed.

But what acutally happens? Is it that these links really are devlaued for their anchor text and the other top phrase search terms for the destination site? Is it an 'upper limit' ceiling which is imposed on the site, it can't rank above a certain level no matter what they do for a certain period of time?

If it was basically a 'devaluing' of the specfic links, a site owner could get around that by simply working hard to build new links for their brand terms. However, JohnMu's suggestion to file a reconsideration request seems to imply that the whatever has been imposed on the site can be 'lifted'. In other words, they can't get out of the cellar until they clean up their act.

Paid Link Reports in Google

There has been much speculation about what happens to paid link reports that are filed in the Google Webmaster Tools. Eric Enge of StoneTemple just interviewd Maile Ohye, Tech Solutions Engineer from Google who worked on Webmaster Tools. I think this is pretty clear:

Eric Enge: Right, so then next question for paid links that get reported in Webmaster Tools. I think it’s fairly well stated that the primary thing you do with it is use it to improve your algorithms. But, if you do confirm a link is a paid link when you get the report, is it normally disabled from passing PageRank?

Maile Ohye: Yes, we do disable such links from passing PageRank.

and a bonus tidbit from the interview, with Google now indexing some Flash text:

Eric Enge: So, will a link embedded in Flash pass PageRank?

Maile Ohye: Yes, it functions as a regular link.