Google stemming - the new algorithm isn't rocket science!

Do you have a question? Post it now! No Registration Necessary.  Now with pictures!

Threaded View
Google added stemming to it's pre-search processing last month - it's
a RADICAL addition to their arsenal of search processing - as it
selectively stems keyword phrases before applying them to the search
databases - this means that a search for:

denver web hosts

will also yield results that would previously have been:

denver web host
denver web hosting

This is HUGE in it's significance, as it means that a stemmed result
will be different than a non stemmed one.  You can turn off stemming
by using a - qualifier - ie,

denver web hosts -anythingatall

will automatically turn off the stemming - and consequently, yield the
previous results for denver web hosts - ie, with only hosts in the

You can verify this yourself by looking carefully at the highlighted
results keywords.   You will not find it hard to find see that other
words are highlighted than just your original search phrase, UNLESS
you use the negative qualifier.

This is 100% borne out by the change in the basic search help page at


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. 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.

Quite the reverse from the previous version found at -
dated Feb, 07 2003:

Word Variations (Stemming)

To provide the most accurate results, Google does not use "stemming"
or support "wildcard" searches. In other words, Google searches for
exactly the words that you enter in the search box. Searching for
"googl" or "googl*" will not yield "googler" or "googlin". If in
doubt, try both forms: "airline" and "airlines," for instance.

My inclination on the future of stemming at

1. they tried it out
2. it's working - they're extending it
3. they originally stemmed singulars and plurals of most words, plus a
FEW very popular terms - more will follow.
4. they will roll stemming out across lots more phrases, starting at
the most popular moving to less popular over the coming months

If you think about it logically, previously, has applied
changes in steps - see if it works, roll it back if not, roll it out
if it does work.

The fact that not all searches have been stemmed where possible, leads
me to think that SOME of the phrases currently being stemmed have led
many experts, to come to this conclusion about "commercial phrases"
being messed with - if you subscribed to this theory, you're right in
a way - but it's because those phrases were popular - they hit the
first round of stemming - ie, the pre-search filters for popular
phrases have stemmed, thus they hit different pages than previously.

The 2nd update in about 3 weeks and the fact that it is still working
(the stemming), tells ME that it's been a success as far as they are

I'm convinced the stemming will get more and more - you will see more
and more people complain about lost SERPs in the near future.  The
problem "stems" from the fact that SEOs have been optimizing for
"exact phrases" not complete topics - ie, an SEO that decided to make
a page rank for a phrase will do much worse in the new google than one
who looks at a broad range of phrases and applies them across range of
pages, or sections of a site, or even a whole site.

The days of the google free ride are over - content and the ability to
rank well for a whole topic are the way forward - SEOs are trembling
because they are grabbing onto straws such as your articles, and they
are NOT finding the answers they need.  I can't offer them much help
unless they are our customers, but I have to say, our ranking for our
own site have NOT dropped, as we had a very broad strategy - I'm glad
to say we weren't as short sighted as many.

I'm sure you'll find this article useful - good luck!

Greg Hewitt-Long
-- /
$10 domains at Web Your Business!

Re: Google stemming - the new algorithm isn't rocket science!

that is what i figured...

however... the question is also whether there's a deliberate stemming filter
meaning that if a site as an *unnatural* profile of backlinks and such it
drops them significantly...  for instance...if ur company is a 'web host'
people will naturally link all sorts of things like:

#backlinks / keyword
100 / web host
90 / web hosting
80 / hosting
85 / web hosting company
150 /
50 / hosting servers
68 / host web sites
15 / our host

so in the case of SEO's they have done one keyword or two...such as 'web
host'... this means that their backlinks look like

#backlinks / keyword
100 / web host
9 / web hosting
3 / hosting
5 / web hosting company
15 /
5 / hosting servers
6 / host web sites
1 / our host

hypothetically of course...
i'm pretty sure this is going exactly how they are determining
quantitatively is the real question...

this explains why some industry organisations have dropped to oblivion...
for instance...if there's an association called 'Paris Hotel Association'...
people would generally link to them with that exact anchor text... and if
there's too much of that they may be dropped for unnatural linking...

anyone proved or disproved this?

Quoted text here. Click to load it
Quoted text here. Click to load it

Re: Google stemming - the new algorithm isn't rocket science!

So in order to maintain a high reanking we need to add plurals?
Example: Place Mat
Use place mat
       place mats
       places mat
       places mats

Re: Google stemming - the new algorithm isn't rocket science!

On 8 Dec 2003 10:09:19 -0800,
(Greg Hewitt-Long) wrote:

Quoted text here. Click to load it

IMO stemming makes optimising a site much easier.

Take the hosting example.

We can now fill a page with related phrases host, hosting, hosts
etc... and they will all count towards the same SERPs. So in one page
you can optimise 2, 3 or even 4x as many phrases and use related words
to boost your main SERP!

This is great as it will make it easier to add keywords without making
it look spammy, more choice in keyword for the same result.

The smart thing to do is pick one of the phrases, optimise for it, but
add the related phrases as many times as you can as they count. With a
little luck you'll get several of them.

I run a commerce site and a big problem for me is when I describe a
product you tend to use the singular form of the word (plural is
harder to use), but with some SERPs the plural is used far, far more.
Stemming will help a lot with this so I'll not have to work as hard at
getting the plural versions in the copy.

Overall stemming is a good thing for SEO.

I realise stemming hasn't been fully rolled out yet, but I've seen an
increase in traffic since the 16th. Before the 16th I got 5500 unique
visitors a day on average, now I see days with over 7000.

I've not checked to see if stemming has anything to do with the
increase, but will when I get the time.

Free Search Engine Optimization, SEO and
Search Engine Placement Tips (updated 16/11/2003) /
Free SEO Mailing List

Re: Google stemming - the new algorithm isn't rocket science!

many of the people who were hurt by the changes were not necissarily
short sited, they were just doing what worked well. as things change
so can we. there will always be ways to exploit search engines, just
over time it will get harder. an important lesson is that people
should learn better to create lifetime customers vice going for 1 time
free marketing
aaron wall

Quoted text here. Click to load it

Re: Google stemming - the new algorithm isn't rocket science!

I have a theory. If the stemming words all count as the same words, doesn't
Google become more sensitive for keyword stuffing then? Because if someone
has the words, let's say "walk" and "walking", on the site, and has 10 occurences
of walk and 9 of walking. Then the keyword density would increase (almost double)
after Florida, because they count as the same? And with a too high keyword density
sites might have got penalties, which might explain why they've got a worse position?
In particular sites, which have been on the edge of having a too big keyword density.

Am I wrong or?


Re: Google stemming - the new algorithm isn't rocket science!

Quoted text here. Click to load it

Filter Shmilter - I pasted my keywords on my home page and wouldn't
you know it - my top rankings returned.   While this may only be a
short-term solution at least my new business generation has returned
for the moment.

Re: Google stemming - the new algorithm isn't rocket science!

Quoted text here. Click to load it

I would say the total opposite.
If you have been concentrating on one key word
example "walk" there may be thousands of other sites
which have "walking" on there site, which could all be
pushing your site down.

Re: Google stemming - the new algorithm isn't rocket science!

A question... We have all noticed that google marks the words in
bold.. but it does not show them up in the cache as highlighted? Do
you really think that google is counting all words as the same?

I have gained my page 1 rankings for

horse sales #4
horses sales #2
horses sale #1
"Horses for sale" #4
horse for sale #10

but I'm still page 3 down from page 1 before the 6/11 change over for
the term:
horses for sale.

Either way like Dave theres been no significant drop in traffic during
any of this due to the vast varity of search terms entered. This
particalar site tends to cater for people who are very speific in what
they are looking for.

Stemming may make things easier for some but I havent figured out how.


Site Timeline