25 replies to this topic

[bwelford]

Google now uses stemming technology. Thus, when appropriate, it will search not only for your search terms, but also for words that are similar to some or all of those terms. If you search for "pet lemur dietary needs", Google will also search for "pet lemur diet needs", and other related variations of your terms. Any variants of your terms that were searched for will be highlighted in the snippet of text accompanying each result.

This is taken from the Google Search Tips. Until recently it said quite the reverse. That's a pretty fundamental change to bring in without a fanfare! :glasses:

[Black_Knight]

Nicely pointed out, Barry, and yes, it is a pretty fundamental change. It could be quite an annoying one for those who've carefully practiced 'niche SEO' - finding the search term variants that were less competitive.

Previously a page optimised for the phrase "online gambling' (purely for an example) was competing with pages that also ranked well for that term. Now a page optimised for "online gambling" is also competing against pages optimised for "gamble online" and maybe even "online betting", "bet online" etc, etc.

That's a massive change for SEO to take on. It will have both positive and negative effects for us. For Instance: it may allow a site that dominates the SERPs for popular phrases to automatically have the same dominence over less popular phrases without any additional work. Whether that is good or bad will depend on whether that dominating site is our own, or a competitor we hoped to 'out-niche'.

The LocalRank filter also seems to have come into effect without any great fanfare, and while we suspect it has been in the general algo mix for a while, it seems to be more notable since this big shift in algo and filtering.

In fact, it could be that we are not seeing a single change (which many are going mad trying to decode and assign motive to) but rather a whole mess of smaller independantly motivated changes that have simply coincided and thus made it far harder to see what has changed and why.

[Jean_Manco]

[Edited to remove irrelevant post.
Sorry - I was mixing up stemming with the "Related searches" feature that Google has apparently been experimenting with since September. ]

[peter_d]

Yes, full details here: http://www.google.com/help/basics.html

Out: keyword selection
In: semantics

[sanity]

Great find Barry thanks. What a week this has been eh. I believe Google has changed the playing field.

[samsagaz]

can someone made an resume of all the new improvent and changes of google?

[peter_d]

can someone made an resume of all the new improvent and changes of google?

Yes, but they work for Google The rest of us hypothesise....

[projectphp]

can someone made an resume of all the new improvent and changes of google?

Yes, but they work for Google The rest of us hypothesise....

ROFLMAO LOL LOL

Peter, your sarcasm and ability to cut through the BS is BRILIANT!!! Are you sure you aren't Australian?

[DaveChild]

Chris has come up with an interesting idea based in part upon what you spotted, Barry. It's an interesting read, and it does sound to me like he's hit the nail on the head ... http://www.searchgui...13-0.html?54529

[ricka]

So here's the synopsis of his theory:

So my theory is: there is no filter, just badly implemented stemming.

How does that account for my wife's site (http://book-cover-design.com) going from #1 to nowhere for the term "book cover design"? She still has good page rank and still shows up #1 if you override the filter.

[tosheroon]

Thats the fourth time you've brought your wifes site up.

I'm amazed the site was ever number one. The home page is all images even for the title "book cover design", the contact info and the links - the phrase book cover design only appears in the title, description and keywords never in the visible text. There's a few inbound links showing mostly from your site.

For a filtered search The links page is the first to show. The DMOZ and Google directories list it without the www.

I think you'd be better off adding some text and doing some optimization on the page before you complain about google. :evil:

[ricka]

Since you're counting, I won't bring it up again. I think it did well based upon the key word in the domain name, which automatically placed the keyword in the link text from other sites. That and the title tag. I didn't bother doing anything more to it since was ranking well as is. But now I will. Thanks for your feedback.

[tosheroon]

sorry for the rant - too much coffee today.
but this might indicate that a major factor in the update is that much less power given to keywords in URL.
which is of course obviously done for seo if its multiple keywords, without seo you would probably have gone with archer design as the description would tell the public what the site is about and the title would be used for branding.

[markus]

I've noticed for some searches the rankings pretty much followed allinanchor keyworda keywordb. Now for some of these results i'm seeing the rankings listed by, the average position the site has for allinanchor keyword a and allinanchorkeywordb maybe just chance, but i doubt it.

[compar]

but this might indicate that a major factor in the update is that much less power given to keywords in URL

I've seen other people suggest that but in the last couple of days I have started to suspect the opposite. I've seen some very mediocre sites from a traditional SEO point of view come up #1 and the only possible explanation seems to be the KWs in the URL.

Check out this search for instance. <http://www.google.co...line pharmacy>.

Look at the sites at #3 & #4. BTW number 3 was at number one yesterday so adjustments are still going on.

<added Dec 1.> The site that was in #3 when I posted the above has now returned to #1 position. I believe that a careful examination of the information available at Scroogle will reveal exactly why this site has survived the shake up and the filters.<end of edit>

[compar]

We seem to have taken this thread off it's original "stemming" topic, but here is soomething else I've noticed.

We all know now that you can use the advanced search "-junkword" and get virtually the same results for you keywords as existed before all the machinations.

However I have never heard anybody talk about the impact of doing searches enclosed in "". I have found that for some -- not all -- of the keyword phrase that have disappeared this search will restore the site to approximately it's former ranking.

Does anybody have any theories or explanation for this phenomenon?

[bwelford]

I would encourage everyone to look at their website traffic logs and in particular those visitors who arrive via Google searches. I was pleasantly surprised at what I saw. The new Google stemming seems to be bringing visitors who searched for a related word and ended up on a web page that did not contain the exact word but a "stemmed" version of it.

My guess is that if you have not been penalized by the Florida update, then your web pages may appear on more SERP's for stemmed versions of your important keywords. In turn that is likely to lead to slightly more traffic.

Has anyone else noticed this phenomenon?

[DianeV]

Well, not exactly. But here's an example of the new! improved! SERPs! that I saw over at SEF.

Note the one site with three listings on the first page. Then start clicking to the next pages.

Google must be proud. While this doesn't prove stemming, it does show exact phrase matching. Only if you're an "authority", I guess.

Er, back to the topic at hand.

[compar]

Barry,

That maybe the case for web sites that did not trip the filters, but it is my feeling that if your keyword phrase is on the hit list any stemming for that term has also been turned off.

[bwelford]

I've got to amend what I said previously. You may have read the following from the Google Help web page on Basic Search.

Word Variations (Stemming)  
Google now uses stemming technology. Thus, when appropriate, it will search not only for your search terms, but also for words that are similar to some or all of those terms.

Well it isn't quite true. What Google is doing is much more clever than stemming but it isn't stemming.

In what follows, I am assuming that the highlighting feature on the Google Toolbar correctly identifies how Google is treating the words in its Search Algorithm. If Google was really stemming then a search for Friesian Horses would highlight Friesian and Horses in the composite word FriesianHorses. If you have the Google toolbar, you can check it out in this search by clicking on the highlight button.

This shows that Google found the word Friesian in Friesian-Horses, in Friesian_Horses and even in Friesian+Horse. However it didn't find Friesian in the composite FriesianHorses.

So Google I still congratulate you on what you are doing but it isn't really stemming.