Forum Moderators: open
[webmasterworld.com...]
Yet, I think the subject is important. Here's the issue: Google claims that it ignores common words in searches. It certainly *used* to. But it doesn't seem to anymore.
For example, try a search on "widgets in placename" (without the quotation marks). Now try a search for "widget placename" (again, without the quotation marks).
Google will say: "in" is a very common word and was not included in your search.
If that's the case, then why are the SERPs so different for each of the searches?!
If the common word was ignored, the SERPs should be the same for both. As far as I recall, this used to be the case. Not anymore however.
Is Google misleading searchers?
e.g. position one (widget) = 0.5, position two (in) = 0.3, position three (placename) = 0.2 etc.
So although "in" has been exlcuded from the matching process, its position still had some weighting.
For example, try a search on "widgets in placename" (without the quotation marks). Now try a search for "widget placename" (again, without the quotation marks).Google will say: "in" is a very common word and was not included in your search.
If that's the case, then why are the SERPs so different for each of the searches?!
I must be missing something here. What you describe seems to be normal Google behaviour.
In searches with very common words, yes, the common words are ignored, but the word pattern is not.
Thus, the search:
widgets in placename
Becomes:
widgets someword placename
Which is not the same as:
widgets placename
Different search phrase, different results, normal Google behaviour.
widgets in placename [google.com]
and
widgets of placename [google.com]
- seems to me that the 3rd, 4th and 5th results are swapped in the two searches. Most mysterious...
'widgets placename' says to boost pages that have widgest right next to placename, but 'widgets in placename' says to boost pages that have widgets within one word of placename.
I assume that Google doesn't index the stopwords, but it may use some of them in queries. I tried using "to" instead of "in", just in case "in" is special. Got different results yet... but, as with "of", no Local results at the top.
[edited by: Robert_Charlton at 5:01 pm (utc) on June 27, 2004]
widgets in placename 233 results
widgets placename 234
widgets of placename 233
widgets from placename 233
widgets where placename 233
widgets why placename 233
widgets on placename 233
I think Hennatron's explanation probably sums it up.
placename widgets in 227
placename widgets 233
placename widgets of 229
widgets of on in from where why how for placename 234
widgets of OR on OR in OR from placename 233
widgets of AND on AND in AND from placename 235
widgets NOT of NOT on NOT in NOT from placename 217
Well, maybe my mind is changed. Here's some evidence which you can gather using the boolean NOT. If Google isn't searching for these words anyway why should this make a difference?
microsoft 102,000,000
microsoft NOT of NOT on NOT in NOT from 10,200,000
microsoft -of -on -in -from 5,450,000
microsoft +of +on +in +from 10,700,000
This is starting to look really funny. I thought NOT and - had the same effect? What's going on here?
I have always had success optimizing for "Widgets in Cityname" separately from "Widgets Cityname" and separate still from "Cityname Widgets". I don't ever remember having success for all three variants with one page.
"in" triggers Local results at the top of the serps
That could be a good explanation for many of the discrepancies.
Could it have something to do with weighting applied to the position of each k/w in the search string?
There should be no difference, they should return the same results.
I'm intrigued by Robert Charlton's suggestion of localisation for "in".
it used to be that
widgets in placename
widgets to placename
and
widgets * placename
would all give the same results. Now we're seeing differences. Good to see.
For literally years I've been trying to get my "how to perform widgeting" page to rank for a how to perform widgeting search (the four words, no quotes). Unfortunately google has always ignored "how to", and thus ranked a stronger (but less relevant) page higher, based only on the "perform widgeting" part of the search.
Several days before the recent backlink update my "how to" page started outranking the more generic page for that "how to" search. Google still says it is ignoring "how to" (and not highlighting it in the ransom notes) but perhaps there is some common word change being implemented or experimented with.
<The "in" thing is a fascinating idea.>
As a side note the term "in" appears to be only part of the local formula. For local results to appear, "in" and a geographic term need to be part the search phrase. I've also seen local results without including "in" and sometimes I've seen no local results with just the geographic term and not including "in".
Google says: "The following words are very common and were not included in your search: where is"
I'm seeing a different result, even when Google says they ignored the words, compared to me leaving the 'where is' out.
Do a search for:
'Something is' -
[google.com...]
Google says that it takes out 'is' because it is a common word. So the results should be the same as a search for 'something':
[google.com...]
But they're completly different. So I was stuck for a while not knowing which phrase to focus on (the above are examples only), in the end I gave up trying to figure Google out and optimised for both... :)
Is Google misleading searchers?
"widgets of usa"
"widgets usa"
althought google would say that it ignores "of"
This is nothing new either. I've been noticing it for almost a year now.
So Google ignores the stopwords, but not their effects on proximity? (Has this always been the case? I'm interested that, like me, many people have only noticed this within the last year or so -- maybe since Florida?)
Michael, stopwords like "in" and "the" are not used to filter the results (at least not when they're treated as stopwords) but the proximity of the terms in the search is counted.
hotel in mars, mars hotel and hotel mars
0........9..., 0....5.... and 0.....5...The proximity from hotel to mars might be indicated as +9, -5, and +5. If the text matches the query then it's a much better match. For phrases where the top listings are quite close in other respects, this proximity can affect rankings considerably.
Ciml, I'm quoting your message from one of the threads you linked to, as I think it's a neat explanation of how "proximity" works.
And yes, some initial searches I've done produce identical results when two-letter stopwords are interchanged (widgets in placename = widgets of placename), while different but, again, consistent SERPs appear for three-letter stopwords (widgets for placename = widgets the placename).
Robert -- interesting point about the "in" producing local searches, and being different to "of", which would indicate an anomaly. I'm not seeing it in my own tests, but it may be worthy of further investigation...
Goes sometimes with the "good grammar" version as higher placed in serps ..and sometimes not ( seems to depend on the number of letters in a word ..some common words have 4 or 5 letters and in some cases with phrases the words can include punctuation symbols making them longer )!
I know this isn't down to "offpage" cos I have run the experiments myself using my own pages and "optimising" separately for each scenario pages which are otherwise identical linkwise etc ...
Question is is this filtering behaviour geo specific or language specific ..any one else seen evidence in non English serps?
A search for: a state
Returns 266,000,000 results and "a" is a very common word. You'll not see a bold a in the first page of snippets. It's ignored.
A search for: a-state
Returns 6,940,00 results. There's no mention fo common words. In the snippets you'll see "a state" in bold. That's "a" space "state". You get seperate dictionary links to "a" and "state".
In the a-state search you're getting "a" as a seperate word and counting it - but only, it seems, if it's right next to "state".
Perhaps Google should say, "a is a conditionally common word"?
Hmm. Contextually common word?
I do not hope for Google that they deliver many results like this. If you were to search "Donald Duck drank tea" in English would you be satisfied with a SERP that did not count "Duck" and "tea"? And would you like having an explanation that showed complete ignorance of the English language?
At the present quality level Google's common-word-filter in Danish is a nuisance.
...interesting point about the "in" producing local searches, and being different to "of", which would indicate an anomaly. I'm not seeing it in my own tests, but it may be worthy of further investigation...
I hope I can post this without violating TOS... it's one of those generic type searches that is nothing more than an example...
hotels in Omaha
hotels of Omaha
hotels to Omaha
The same thing happens with other city names. Not sure what might happen with other widgets.
