why, why, why, why, why?

[hawkwind dave]

Why, why, why, why, why does google do what it does?

I've just been checking the SERPs, and noticed this:

in Google.co.uk, 'flowers for funeral' shows my client at number 7. However 'flowers for a funeral' shows my client at number 4.

and at the top of the page google says "The following words are very common and were not included in your search: for a"!!!

so if 'for' and 'a' are not being searched for, sould the results for 'flowers for funeral' & 'flowers for a funeral' be the same?

[Faglork]

Just to make it clear: Did you search for the PHRASE, which means INCLUDING quotationmarks, or for the WORDS, i.e. without quotationmarks?

[hawkwind dave]

Without quotes.

dave

[tcampione]

Because stop words really do seem to have an effect, even though GG says they don't.

Our company is in the rental business, primarily.

When I search for: ourMainProduct rent we are #1. When I search for: ourMainProduct for rent were are #3.

It's been like that for the past 9 months that I know of. The same illogic goes with singular and plural version.

There seems to be greater parity now than before the Feb 7 update, but that statement is based on only the few keywords I follow.

[MathsIsFun]

Just a theory, but for any search engine the speed of lookup is important, so you try to cut your "search space" down as quickly as possible.

So when you ask for :flowers for a funeral:, you would start with 8 billion possibles, then cut down to "flowers", then cut that down to "funeral", after that you may not cut down any further, because nearly every page would have "for" and "a" in it.

Those results are then handed to the "score 'em and show 'em" routine, which may play around with the "for" and "a" terms a bit.

So, I think Google may be saying "we haven't included for and a in looking this up, but we have used them to tweak a little"

[cyanide]

lol Dave,

I've been wondering about that for a while too.
Lemme know when Google tells you all about it

[kservik]

I have seen the same thing.

[DrTandem1]

Yes, articles such as "the" and "a" are supposedly dropped from the search. Obviously, they do have an effect, albeit slight. Try dropping the preposition (for) as well.

I think what Google is saying is that those are not important to the search. However, with their ever increasing sophistication, they are learning to incorporate such words as modifiers to the search to bring about the most human results possible.

The statement that they don't use them is probably more a throw-back to when they actually didn't use them. After all, what would a search for a word such as the article "the" return? As it happens, the US government's White House is #1. Makes sense. No one ever talks about "a" white house. It is so much a part of the news that it has to be a common anchor text. That is your clue. The "unimportant" word is important, if it is used in as anchor text.

I hope that helps explain the phenomenon.

[martindow]

Could this effect not be due to Google giving weight to words due to their position? That is the first word is considered more important than the next word and so on.

Google may well be not using stop words in its search but is noting the position of each word in the search string. If you include a stop word in a search then you are moving subsequent words further along and reducing the weight given to them.

[Mel]

I suspect that Mathisfun's answer is very close, but IMO the actual process is slightly different. When Google parses a page it puts each word it finds in an "inverted index". For instance Google puts every page with the word funeral occuring on it in the funeral pigeonhole together with the associated data for that page.

When someone asks it for pages with the search term flowers for funeral, it returns a list of every page that has each of these words on it, and this is the list (ranking pool) that is handed to the ranking algo.

When someone ask for the search term flowers for a funeral it returns a different list of pages to the ranking pool since it has added a bunch of pages with the word a on them to the ranking pool. So now the ranking pool is different and the rankings from that pool are different also.

In this case if you use the phrase flowers for a funeral on your page you can expect to rank better for that term than for flowers for funeral and vice versa.

[AccuraCast]

I think Google tries to match the exact phrase (including stop words) as far as possible, even though it claims to have dropped the stop words. What probably happens is that the first few results we see includes the exact phrase, but as it ploughs on, it finds more results without the exact phrase, in which case it then takes the liberty of including those without the stop words.