homepage Welcome to WebmasterWorld Guest from
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Become a Pro Member
Home / Forums Index / Google / Google News Archive
Forum Library, Charter, Moderator: open

Google News Archive Forum

This 143 message thread spans 5 pages: 143 ( [1] 2 3 4 5 > >     
SERIOUS Brandy update algo analysis thread.
NO whining or cheering about how your site is doing in this one.

 11:54 pm on Feb 23, 2004 (gmt 0)

I say that it looks like this update is done, and this thread is now due. For those unfamiliar with the history of these "SERIOUS" threads by me, the ground rules are:

"This thread is ONLY for serious, generic discussion of changes that you are observing with the new algo in this update. As in things like "Looks to me like PR is less important this month, and anchor text of inbound links counts more.", etc. How your site is doing has no relevance here unless you can explain why you think so in terms of a general algo analysis."

This thread is NOT for y'all to say how much you think Google sucks, or alternatively how great the new SERPs are. The idea is to pick apart how Google is working, and not to criticize their quality.



 1:22 am on Feb 24, 2004 (gmt 0)

A major improvement I've noticed is the reduction in interlinked or closely related sites dominating the first page or two of any particular serps.

In many areas I monitor, any set of intereconnected sites are now reduced to a single listing in a serp whereas they used to have three or four.


 1:49 am on Feb 24, 2004 (gmt 0)


You are exactly right. Due to strong brand differences that we represent we devote a different site to each brand and they are interconnected. Under different circumstances this could be considered spam but in our case it is just good buisness practice. Of couse the different sites are what is currently being classed as overtoptimized for the particular brand names although we and the other search engines still consider it good practice. Keep in mind that we do not spam at all. Everything is super clean including the cleanest possible code. It is my beleif that our sites have suffered radically for these two reasons. From dominating our money words to totally gones for all the most important ones and yet still ranking well for the less obvious ones.
This is what leads me to beleive that we were not penalized but filtered on certain keywords. I would imagine that a list of manufacturers could have been part of the dictionary used.
In my particular case I am not willing to give up Y and MSN to chase Google so we are sitting tight untill threads such as this one can start to make some sense of things.
And watch our adwords bill go up.


 1:59 am on Feb 24, 2004 (gmt 0)

One change I possibly see is a greater importance on internal anchor text than post-Florida. My own site, the only one I'm basing this on, seems to be passing PR, or anchor relevance that boosts serps, more like what we had prior to November.

This is based on the observation that one of our obvious kw searches, that put www.oursite.org/main.htm at #1 from Sept 2002 until Florida, dropped to #2 behind an "About" page, (that has five links to our site on it), with Florida. /main.htm has now gone back to #1 with a supp listing, the index.htm as #2. The About page is #3. /main.htm has anchor text of the particular kw-kw on about 80% of the internal links... it's the main contents page and every other page on the site, 200+ pages, links back to it.


 2:13 am on Feb 24, 2004 (gmt 0)

My experience with "Florida" (besides highly suspect vote counts in Election Year 2000) and now Brandy has been the following:

The clients I work with took a hit on their main 2-word money phrases...for one of my client's he is dead set on a very narrow set of keyword phrases for his two main properties because he knows that these are the ones that convert to sales...I have always been of the conviction that you need horizontal keyword phrase distribution..instead of vertical...something I learned in the early days (1996) in the SEO business..

You want to spread your keyword assets out and reach the broadest and deepest audience sets this way..

Naturally, we switched to some PPC to pick up the slack as I worked to bring the sites in alignment with the new word stemming algo integration...I was starting to see some positive results of my efforts when Yahoo disconnected and Google decided to revert back to a previous index set (or at least implement a different blend of their new dual algo to bring back index pages for 2 word phrases...perhaps the recent increase in AdWords revenue has allowed them to take a breather from their new direction...I do not believe that the current index, such as it is, will survive)..
I believe Google took some of the treasured DataCenters offline and kept the old SERPS in a preserved state for the last several months and then brought them across live without any of us inquisitive types being able to take a looksee...

With the current SERPs our sites are back just about where they were prior to Florida (but the votes where never recounted properly...shame)...so we are dipping into the sweet nectar of natural SERPs traffic, though somewhat reduced, with Yahoo taking something in the order of 20% of the search market to their table...is that correct?

We are currently doing nicely with Yahoo and MSN so for now we have reached a reasonable balance...but I have informed my clients ... be ready to crank those PPC accounts again when the next Google update comes along...it could be uguuully...

On The Technical Side
...I believe that one of the things Google is trying to initiate is a more organic link structure....in the same way the hypertext documents were strung together before the WWW Browser became prominent (prior to 1994..you know Marc Andreesen? and Netscape)...this way link relationships flow more naturally from site to site...case in point...look at how many links Yahoo lost in the last update...from near 800,000 to 490,000 ... this indicates a scale down of centralized link pointing to more of a flow-through model...where the theme weight is carried from site to site...and the qualification and quantification of a pages importance is calculated in this flow pattern...just a thought...tell me to jump off a cliff if you don't like...I'll just strap on my parachute and have a blast...

Another factor to consider is the word stemming issue...this certainly tends to favor a more generic view of a page then the very specific on page optimization of the past...so one needs to be able to broaden their dialogue but still have it make sense for the visitors (this is where the sales come from...unless the engines are planning on making the bots "purchase agents")
So to accomplish this one needs to start at the very top of their respective category for their particular keyword structure and find a practical, elegant way to integrate some of the primary variants into the copy...not an easy thing to do and still read well...

Thanks for listening...
Good luck to all...now back to work making money for my clients...

[edited by: decaff at 2:31 am (utc) on Feb. 24, 2004]


 2:28 am on Feb 24, 2004 (gmt 0)

I have a very similar situation to BouyBoy. We have our main site with all products and then sub-sets of products listed in specific manufacturer sites based on our agreements with those manufacturers. In most cases (prior to Brandy) we dominated the top rankings.

Now I see only one site in each keyword group from our "network". It's not always the same site. Sometimes it's the main site and sometimes one of the manufacturer sites. Are they filtering based on IP block or is it this "Latent Semantic Indexing" I am reading about? The basic product info is identical from one site to another.

I'm still reviewing this for now.


 2:48 am on Feb 24, 2004 (gmt 0)

One interesting thing I have noticed in my (very popular) area is that page size now seems to be more of an influence.

In general:
Pages < ~10K or including indexes with framesets are being pushed down
Pages > 10K and < 35K are getting a boost (or are just boosted by other sites going down)
Pages > 35K are getting pushed down

This is consistent with the recent efforts to reduce SPAM, promote faster loading pages with content, and not overwhelm users at 56K speeds.

This is based on a comparison with pervious SERPs.


 2:49 am on Feb 24, 2004 (gmt 0)

My site's been bounced back from #2 to #5 and #4 to #8 on two of my most relevant searches. The sites ahead of me are definitely greater authority sites than mine, with more inbound links from on topic sites. I used to kick their butts because of my heavy internal linking from the number of pages I've got and the greater number of on-topic pages. I think the SERPs are better now -- though one of the authority sites seems to have pretty old, incomplete information.

Also, I used to do well on a page "black widgets, white widgets, big and small", but after brandy, my site was filtered out of SERPs for "white widgets" but not "black widgets". White widget is a much more competitive phrase. I think that the repeat of "widgets" did me in on that search. But when a third keyword was added, such as "big white widgets" I am #1 -- so the filter dropped on third keyword.


 2:53 am on Feb 24, 2004 (gmt 0)

Ill preface everything below by saying that I am more convinced that Latent Semantic indexing is the new toy in the plex.

[webmasterworld.com...] .

Observation 1. Anchor Text more weight:

Link from :

My site . Five pages with PR4 link out with a three word link.

Link to :

Site with related keyword in two word title .. second word is "web" ( ie irrelevant. )

Site with PR3.

Keyword density of Keywords light .

On topic body ( relevant to the KWS i link with ) .


Beats PR4 site which kills itself trying to rank , but with little content.

Observation 2 Giving away pr to on topic sites .

My site:


Body is relevant to keywords and includes keywords.

Outbound links to both subdomains and sites which are on topic .

No inbound kw links.


PR sites which have title , tags and KWs in text . with inbound links with KW's pushed into 4th and 5th place.

Think there was some people talking about an over optimisation penalty a while ago, and im sure they do exist... however could it be that the sites are not considered topical .?

Observation 3 Anchor Text ..plenty of weight

It stands to reason that organic pages will include topical keywords in text and anchor text.


Sites with density of anchor text that contain KWS rank well. However one site that I have seen rise on this premise links out to topical sites. Im going to play with this with a playground domain and see if i can see the difference between linking to on and off topic pages.

Personally given the raw state of LSI i think google should turn back this knob a bit. Most surfers dont know how how to search a web page to find out where their KW is...which is what i have found myself doing.

DMOZ and Directory Listings

Perhaps old news but new to me. But if you can manage to pass your kws into the description in DMOZ extra weight seems to be given to this beyond on page factors. Ive seen ( and griped about ) a listing that just should not be where it is based on DMOZ listing alone.

Bottom Line . On topic links in and out , with on topic content and interlinking .. almost sounds organic. Not there yet but going in the right direction IMHO. I am more and more confident that i have to spend less time here and more time writing content for humans that humans will want to link to.



 3:04 am on Feb 24, 2004 (gmt 0)

Internal anchor text carriers greater weight. I restructured my internal linking text and it made a HUGE difference.


 3:27 am on Feb 24, 2004 (gmt 0)

Wellzey wrote:
>>Internal anchor text carriers greater weight. I restructured my internal linking text and it made a HUGE difference.

So if you restructured your internal linking text how can you tell that internal anchor text carries greater weight? Maybe it carries the same weight but it is your changed link structure that is the variable that has made the difference, not the algo.


 3:32 am on Feb 24, 2004 (gmt 0)

I restructured my internal linking text

I read that as a change in the anchor text rather than the linking...

Which was it Wellzy?


 3:32 am on Feb 24, 2004 (gmt 0)

Brandy had three parts:
Immediate Brandy
64 Brandy
64-never-really-moved-over-current Brandy

These all had different key aspects to them.

Immediate Brandy and 64 Brandy were pretty interesting. Now, except for successfully at least partially resurrecting many "lost" sites, current Brandy is mostly like it never occured. Clearly extrenal anchor text dependent trivial sites are doing better than the beginning of Brandy and on 64, but there is no major news in that.

The authoritativeness knob has been turned down. More lightweight sites are doing better. This may be the result though of the synonym/semantic knob being turned WAY down.... or at perhaps being altered in some way.

During Immediate Brandy and 64 Brandy searches for:
word1 ~word1
word1 word1
returned excellent results, including some extremely authoritative domains that are nowhere in the normal serps because they have basically horrible almost anti-seo site design. The results for these searches now are much, much more lightweight.

Considering the hubbub, nothing lasting really occured except Google added a lot of trivial and/or fresh piffle to what had been solid results.


 3:41 am on Feb 24, 2004 (gmt 0)

Stefan wrote:
>I read that as a change in the anchor text rather than the linking...

This doesn't affect what I wrote. My point is that Wellzy made the changes, then he claims that after the changes there was a big difference, yet he is saying that the changed algo is responsible for the big difference, not the changes he made. How can he know or even guess such a thing?

This is not an important matter, I just wanted to point it out.


 3:46 am on Feb 24, 2004 (gmt 0)

One other thing looking around now. I'm seeing a bit of the exact reverse of what has been talked about in a lot of threads as being the way to move forward... main/index pages for niche authority sites are showing for multiword search terms where they are authorities on the main topic, but don't have much to do with the specific query. Like a widget domain ranking well for red widgets even though this index page only slightly deals with red widgets.

In other words, the whole iceberg concept has been turned on its head. Making a deep, content rich website with sections/pages on subtopics is definitely not being rewarded now.

If this holds, its a massive step backwards conceptually considering what Google (and GoogleGuy) have been talking about. The results I see now would encourage a person to put all their eggs in one basket, in one powerhouse index page.

I hope it doesn't hold, since that sounds like an awfully boring SEO future to me!


 3:49 am on Feb 24, 2004 (gmt 0)

I have some location based sites that seem to have been hit in the way Krapulator and BuoyBoy have described.

I noticed a while ago some of them were coming up on the same page for some minor terms that did not include a location in the term. I probably should have thought more seriously about this then and tried to stop this happening.

Now most of them have been hit for all the main search terms such as one of them going from number one for a 3 word term that it is easily the most relevant site for down to 350.

There are about 6 sites that interlink from the index page only and these 6 plus another 4 come off a central hub, with most of the external links coming to the hub and just a couple to some of the spoke sites, so I wonder if it might have had something to do with this type of link structure as well.


 3:50 am on Feb 24, 2004 (gmt 0)

After Brandy update GG said:
>"I can try to give you my personal take. I absolutely agree that both commercial sites and info sites provide value to the user. At different times, a user may want either commercial results because they want to buy something, or they might just be looking for info. We want to return what we think the user is looking for, whether it be commercial or info. Historically, Google often started more as a research tool/informational resource. If you did a search for something like "sex," for example, Google was much more likely to give G-rated informational pages than other search engines.
It's not a no-brainer to find the right balance. I think if you asked a sample of WebmasterWorld folks, they might lean toward returning more commercial results. My hunch is that if you asked regular users, they would want fewer commercial results in their search results than a typical SEO would prefer. As for me personally, as long as the user is happy, then I'm happy. I think that to find the right balance, we get valuable feedback from both regular users and from site owners. We'll keep doing our best to find the balance that makes users the happiest, but we're open to suggestions from both sides of the spectrum about how to make searching better."


Brandy #6 for info sites (for "blue widgets"), today is #13 and tomorrow maybe #23 or Florida #...

Commercial sites with fresh tags push info sites down, and that is a big problem. They have the same content every day, while info sites with fresh tag and fresh content lost first page position because of them. I know for all standard problems with fresh boot, but the question is: "What is really fresh?"
Maybe Google must change their algo in that direction.

I see also commercial sites work fine for singular and plural (for "blue widgets" and for "blue widget") and info sites not, despite both have only plural on site.


 3:51 am on Feb 24, 2004 (gmt 0)

This doesn't affect what I wrote. My point is that Wellzy made the changes, then he claims that after the changes there was a big difference, yet he is saying that the changed algo is responsible for the big difference, not the changes he made. How can he know or even guess such a thing?

True. Point taken.

I made no changes but seemed to see a renewed benefit from the internal anchor text. One case only amounts to anecdotal evidence, of course, so this might not be an observation of a real algo change.


 4:16 am on Feb 24, 2004 (gmt 0)

I don't think it's quite finished yet. I'm still noticing that some of the cached pages of my site are old and bouncing back and forth between new/old. As soon as those are consistent then I'll think it's done.


 4:21 am on Feb 24, 2004 (gmt 0)

I also think there is some minor tinkering going on, still a fair amount of fluctation in our results in any given 12 hour period.

Personally I think KW density made a big comeback on this algo. Looking at screnshots and notes of our positions post and pre-brandy, and then doing a quick kw density check on both sets of results sees a strong rise for those with both good KW density and kw anchor links to your own site. This seems to all but have eliminated stupid reults where a page with a link to you on it would be higher in the serps than you were.

I also think that Google has recalculated PR, and its not showing on the toolbar yet. Our own backlinks were recalculated just before Brandy, but I have yet to see any change in toolbar PR. When this starts showing a few more things might start to make sense.

[edited by: ntrance at 6:49 am (utc) on Feb. 24, 2004]


 4:26 am on Feb 24, 2004 (gmt 0)

title tags can be longer and still have the same effect (sometimes even better) than the exact search phrase...

there seems to be filters on ppc providers eg. sites that use searchfeed have been filtered out

backlinks are the most important aspect at the moment

symantics seems to be playing a good part of the ranking algo as well...

and theming, your site has to be "themed"

[edited by: metagod at 4:29 am (utc) on Feb. 24, 2004]


 4:28 am on Feb 24, 2004 (gmt 0)

I agree with Steve B:

"Making a deep, content rich website with sections/pages on subtopics is definitely not being rewarded now."

I'm #8 now in my main keyword-keyword2 search, whereas I used to be #4. No biggie. But some of the folks ahead of me have only a tenth of the on-topic content I've got -- and a few of them have less external links of good PR than me.

I don't think size matters as much now; and the authoritativeness knob, as Steve B said, has been turned down...


 4:39 am on Feb 24, 2004 (gmt 0)

If it is true that internal (anchor) links have so much weight then I should be in all the top10 of the competitive keywords in my area.

For example:

A site that has 30,000 pages, each page have a 'widget' link pointing to the internal 'widget page' and that page have the right semantics and on target as well for widget and no tricks...

based on what most have said...that page should be at least in page 1 of widget serp. Right?

But, here's the problem, that page is not even in the 1st 10 pages and in fact it is not even in the 1st 20 pages or so.

Anybody like to explain that?

In my observation, I believe that internal links have been devalued which is in fact in favor to smaller sites. If internal links are devalued the smaller site which might have a better content than the bigger site will have better chance in getting to the top of the serp based on other factors.

Devaluing of internal link is just logical and in a way prevent internal manipulation. I've seen so many sites that internal linking structure are based on keywords and you get the sense that it has nothing to do with navigation anymore.

Just look at the above example, anybody with extra cash can replicate the same structure thus the serp becomes who can build the biggest useless site.

However....'external (anchor) links'...are given extra weight. Again, in a way it's logical, this minimizes or even take out webmaster's control on SE manipulation. Done right, could truly bring up highly relevant pages...that is in theory of course.

Issues such as multiple throw-away domains pointing to the master site and issues about spammed blogs, guest books, and forums need to be address otherwise it would skew the ranking in favor of pages that has no real value.

For example:

A prolific spammer going through any spammed sources and pointing it to his/her master site or to the affiliate provider site.

But my greatest mystery is this one page that inspite of the different updates/changes pre-florida, post, and brandy remain at #1 where starting from #2 and down everything is fluid.

What is more intriguing is that it is #1 for a 2 word keyword where those words are pretty common and used by most sites in it's category. A solid example of unshakeable position (if it would last forever).


 4:43 am on Feb 24, 2004 (gmt 0)

>One change I possibly see is a greater importance on internal anchor text than post-Florida.

Seeing the same with interior pages.

>But some of the folks ahead of me have only a tenth of the on-topic content I've got -- and a few of them have less external links of good PR than me.

Sounds like two of my pages that were never well placed before that now reside on page one. <title> and <H1> live again. If these werent my pages, I would be pi##ed.

Still think links and anchor text rule the Google roost. Several service industry template sites now do extremely well because of links/AT.

Just read Net_wizard's post. For those of you Hilltop/circa experts, I wonder if local rank comes into play with the internal links in small sites where the index page does well for several kws.

[edited by: Kirby at 4:47 am (utc) on Feb. 24, 2004]


 4:46 am on Feb 24, 2004 (gmt 0)

"Anybody like to explain that?"

One thing isn't everything. Your question is a non-sequitor.


 4:51 am on Feb 24, 2004 (gmt 0)

One thing isn't everything. Your question is a non-sequitor.

And ... do you have an answer to the "non-sequitor question? Or ... did you just want to use the phrase, "non-sequitor" in a sentence?


 4:56 am on Feb 24, 2004 (gmt 0)

Here are my results after Florida, Austin and Brandy:

Post Florida: #1 kw1 & #3 kw2
Post Austin: all over the map but settled in around #3 kw1 and #3 kw2
Post Brandy: #1 kw1 & #4 kw2

All that to say that I use by the book, no spam SEO. And we have a ton of quality anchor links for kw1 & kw2. Index page is PR5. I have begun increasing kw density in title and on page. As well as some theme pages. Still caching in at same position so far.


 5:09 am on Feb 24, 2004 (gmt 0)

"do you have an answer"

What part of "one thing isn't everything" escapes you? LOL.


 5:22 am on Feb 24, 2004 (gmt 0)

catch_22, have a good read of the title of this discussion


 6:10 am on Feb 24, 2004 (gmt 0)

Isn't there something called Normalization in LSI that prevents big documents from overwhelming the smaller documents on count of keywords? Are we sure the same principle doesn't apply to biggie sites overwhelming smaller ones on count of internal links/anchor text? (Kn.Attn. Net_Wizard)

Anyway, the neighbourhood is as much important as anchor text now ... my 2 cents.


This 143 message thread spans 5 pages: 143 ( [1] 2 3 4 5 > >
Global Options:
 top home search open messages active posts  

Home / Forums Index / Google / Google News Archive
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved