webby2001 - 11:42 am on Sep 3, 2004 (gmt 0)

well, I dont know about a definitive conclusion but I have come to MY own conclusion from the data I've collected and the research I've carried out.

I have several tests on the go but it will take a little time for the changes to be indexed and new IBls / internal links to be counted.
I spent most of yesterday taking a large sample of those terms that have remained stable and a same sized sample of terms that have dropped significantly. I created a large spreadsheet for those that were still high and another for those phrases that got hit hard. The idea being to analse and compare the two and see if one group has something the other doesnt or indeed the other way around. The Excel columns were...

KW phrase ¦ No. Internal links ¦ IBLs ¦ PR ¦ Google Rank ¦ Competing pages ¦ "exact" competing ¦ URL

For the number of IBLs I did not use the Google backward link but alltheweb which shows a more comprehensive backward list. PR was a problem as it is so out of date but the pages have been up for some time so havent changed.

On comparison, there were significant differences between number of ibls going to the specific destination sub-pages. One 'hotels in destination' sub page threw me completely and I thought my linkage theory was blown ou tthe window as it only had 4 internal links yet was in the group that kept their rankings and in fact rose. It was only when I checked IBLs that I noticed from the one IBL it had was thematic, had a PR 6 and the ideal link text. The page ranked top for a fairly popular hotel in keyword phrase.

The main differences though were the number of internal links to the specific pages.

Popular / competitive destination sub pages were doing well with an average of around 200 internal links, page rank 5, and one or more IBLs. Those destinations that dropped significantly had sub 30 internal links, no IBLs and PR 4. Link text was in the same format for the good and poor performing pages, so I counted that out. KWD was roughly the same as were title formats, the heading/bold and other tag factors. I noticed a sliding scale dependant on exact phrase competitiveness with regards to the amount of internal links that seemed to working very well. On very competitive phrases internal linkage was often not enough for top 5. The internal linkage was however high (apart from the anomaly mentioned) between 180/290. Just one or two ibls gave a highly competive term (in millions) a top 5 ranking. For minor pages (destinations) 40-70 internal links with no ibls was sufficient to give the destination sub page a top 3 in most cases. It was significant that for minor phrases that were performing badly the average internal linking was less than 15 (mostly <10) and PR was never over 4.

Steps have now been taken to test my conclusion on those dropped in ranking pages with better thematic internal linking structure and to seek IBLs to go to the specific competitive destination pages (not the homepage).

My own conclusions are basically as follows...

Solution for sub-pages that dropped in ranking
===============================================
Increase internal linkage to around 200 for competive exact phrases (phrase in "") and 40-70 internal links for uncompetitive terms. For competitive terms 1-3 IBLs direct to sub page can make a big difference (obviously with ideal link text) especially if it is from a pr 6 page. For mid competitive terms, internal linkage is often enough as long as it is around the 100-150 mark (this of course relates specifically to travel related pages and phrases)

It makes sense to look at the internal link structure as well with a view to pagerank channeling and making sure there is a good and wide distribution. It used to be said to distribute your pagerank mainly to your most important pages. I would say now its important to have good (at least pr 4) on all sub-pages. Although I do feel this isn't primarily a PR thing, but perhaps a number of links with correct link text thing.

Anyway, a bunch of tests are in place. For others there may well be other reasons / new filters for your drop in rankings. I am pretty confident that for the sites I optimize I know what I have to do and will keep you informed of the progress.
Alan