Interesting addition to an old argument...

*.. having keywords from the post title in the url also can help search engines judge the quality of a page.*
http://www.mattcutts.com/blog/scoble-visiting-the-plex/

Interesting -- I'm sure it will kick off an entire new debate about them. FYI, here are our key past discussions on this topic: Commonly Asked: Keywords & Hyphens In Domains & URLs (http://forums.searchenginewatch.com/showthread.php?t=5496)

This was the part that made my eyes go wider when I read it, too.

To me it only makes sense. When I save a file on my computer, I don't name it 1092309234.pdf. Instead, I might name it SEO_Keyword_Analysis.pdf, just for my own uses. In the WWW where linking is huge, it's going to help to use words instead of numbers. Google wants to follow patterns of (natural) human nature for showing us results, so this would only be logical to me.

On the contrary, I would say inflating your file names with your keywords would not be a good idea (not natural), and might do harm to your rankings.

I put in the keywords for my sites/clients, and it has been effective. Anyone have links to some recent case studies on this?

Very interesting.

I looked into this technique about 3 months ago comparing high ranking pages (not sites) and I did notice that when keywords were used in the file names, pages were ranked higher. I didn't know if it was because of that little thing or maybe it was just a coincidence.

G seems to test a lot of stuff, I know file names were included in allinanchor calculations for a short period a couple of years ago, and I suspect the same thing happened earlier with domain names, which led to the Multi-keyword-domain-name-fest.com we can all remember ....

Note that Google, and Yahoo, highlights any keywords in the URLs in its search results.

For example : http://search.yahoo.com/search?p=changpu+river+park&prssweb=Search&ei=UTF-8&fr=FP-tab-web-t500&x=wrt

I would say a keyword in filename / hostname (i.e. in URL) counts as much as in the title tag - that is, a lot.

I changed my most important page names to include keywords and it did help. In any case, using user-friendly names (and that includes for you the the developer!) is a good idea. However, once links and on content keywords are taken into account (lets give them a 1/3, 1/3 1/3 split in weighting), the overall boost might be around 15% based on adding 1 keyword in URL to 1 in ttle tag.

Steve

I would say a keyword in filename / hostname (i.e. in URL) counts as much as in the title tag - that is, a lot.For MSN I might be inclined to agree with you. But for G and Y, I think you're giving the URL too much weight.

Page title is a foundation for SEO success. In competitive queries, it's difficult to rank on a term if the term isn't in the page title. But you can rank for a term if it's not in your URL. To me, that means URL counts less than the page title.

Hi Pleeker,

Your logic is not quite right - but had me thinkig fora few minutes !

Perhaps you should say "It is difficult to rank well if the keyword is not in the title tag OR URL." So the idea of equal weight still holds true.

There would not be many pages that have keywords in the URL that are not also in the title tag. But there are some I have noticed that seem to rank well (disregarding links, which I can do to some extent given familiarity with the sites in my own area).

URL keywords may have less weight than those in the title tag but certainly they have some weight. My analysis of results in my search area suggests equal weighting.

Steve

Also, from a quality perspective, a meaningful file name suggests a higher quality page. Further, whereas a title or 10 or so words can cover a few bases, a filename has to define succinctly what the page is most about so it is not surprising that algos would give considerable weight to keywords there.

Ok ok, filenames can potentially be very long and subdirecties also brought into play - and though uncommon I have seen evidence that this might also be helpful - but personally feel it could easily transistion from good structure to black hat.

Besides, trying to remember if bc.html is blue cheddar or brown cheese soon becomes a nuisance so just for your own benefit it is preferable to always use meaningful file names. Any programmer will know the value of naming variables this way : )

Steve

Perhaps you should say "It is difficult to rank well if the keyword is not in the title tag OR URL." So the idea of equal weight still holds true.Well, you can say it, but it doesn't represent what I believe is true. :)
My analysis of results in my search area suggests equal weighting.I can't dispute what you've seen. I can only discus what I've seen. And I don't think it's accurate to say having a keyword in the URL is as important as having it in the page title. The three building blocks of a successful page are 1) great, keyword-rich content ... 2) a relevant, keyword-based page title, and ... 3) quality inbound links with relevant anchor text. I don't think you can legitimately say "having the keyword in the URL" is as important as any of those. You're welcome to disagree.a filename has to define succinctly what the page is most about so it is not surprising that algos would give considerable weight to keywords thereIn fact, the presence of URLs such as http://www.keyword-keyword-keyword.com/keyword-keyword-keyword.html has led engines to lessen the importance of the actual URL.Besides, trying to remember if bc.html is blue cheddar or brown cheese soon becomes a nuisance so just for your own benefit it is preferable to always use meaningful file names.I never said it's not preferable. I typically use keywords in URLs to make meaningful file names. I just don't think it's accurate to say that it matters as much as what's in the page title.

Hi Pleeker,

My only solace is that your logic WAS faulty. However, that doesn't mean I was right - so I investigated further.

Conclusion : On Google, there appears to be no weight to keywords in the file name (only domain name). A search for terms appearing only in the filename will not return it.

For Yahoo, it matters, but not as much as being in the title tag. Interestingly Yahoo stores filenames in its index but Google does not.

So there we have it. Using sensible file names is just good sense for its own sake and has little significance for ranking (perhaps a bit for Yahoo and MSN).

Titles could be spammed as easily as filenames so don't see why filenames should not be weighted. But this is how it stands.

Apologies for being so sure (and so wrong) earlier.

Steve

*On Google, there appears to be no weight to keywords in the file name..*

If looking at it as an off-page linkage related factor you're no doubt right Kina, but I read it more as an on-page factor strengthening the title/content...

Something bothered me about my last conclusion from my reaserach so i contineued looking.

Previously I stated that no weight was given to keywords in filenames on Google - but this is wrong - they DO count. My error was due to test case. Keywords will not count if they not also inthe title tag or page content. If they are (usual case) then they do count. And the evidence as i mentioned before suggest equal weighting to inclusion in title tag. In other words, it can make a page ranked say 10 jumpto 8 or 9.

In the next post I hope to detail a bit about my test cases.

steve

A few weeks ago I created a new page called 'great_wall_photo_gallery.html'. The content was basically just photos other than the title.

Horror to find a big error of mine re 'use search terms in your content'. What would likely search term be ? 'great wall photos' of course, not 'great wall photo' - duh!!

The title then was 'Great wall gallery'.

Ranking was way down >100. But for 'great wall gallery' it was great ! But of no use !! The fact that 'great_wall_photo_gallery' was the filename was of no help at all - no results for 'Great wall photo gallery'. hence I though filename did not cout at all.

Howver, include the keywords in the content and it does count.

So I changed title to 'Great wall photos gallery' using the plural. Filename still 'great_wall_photo_gallery.html'. Now, 'photo' will be found as part of 'photos' (but not vice versa).

Search for 'great wall photos' - result : rank 10 in index (this is based on content and new title tag.

Search for 'great wall photo' - result : rank 8 (this is based on content PLUS filename).

If we assume (a big assumption but we have to make some assumptions) that there is a somewhat linear decrease in ranking of sites resulted 1 to 10 then the jump from 10 to 8 can be though of as very roughly a 20% jump. This would suggest that keywords in filenames count about as equally as in title tag. A doubling of keyword count and a 20%ish jump (not a higher jump because obviously links are still the same.

Ok, numbers are a wildish guess but to make the certain conclusion - filenames do count, probably, in my eyes as much as inclusion in title tag, but certainly to some extent (so long as in title or content as well as in filename).

Steve

To anyone thinking about startng a new site - think carefully about your domain name - kewords (search terms) used in your domain name trump use anywhere else (Google and others). This is easy to check out.

So think carefully of your primary search term and name your domain accordingly. It is tip #1.

Steve

Kina, you are making a mistake that a lot of people make. Actually, it's not a computer mistake, only a search engine mistake.

The filename "great_wall.htm" is a nonsense phrase to a search engine. The filename "great-wall.htm" contains the keywords "great" and "wall". For a search engine, the underscore character "_" is treated as a hard character.

Why? Because search engine programmers are, uh, programmers. Many coding commands (especially in C) use underscores as part of the command. If a C programmer went looking for help and the search engine interpreted the underscore as a space (which is does for a dash) then they would not find what they are looking for. It's a bit self centered, but that's how it is.

Rename those files using a dash and re-try your experiment - you'll end up with different results.

For the record, when I did my keyword misspelling (http://forums.searchenginewatch.com/showpost.php?p=50150&postcount=18) tests, it was clear that the search engines ranked pages where the keyword was only in the filename. I did not test how well it worked against the title tag, but I suppose you could test the pages and check.

Ian

There is a caveat to lat #1 tip - which may be a big downer, depending on your field. Dependent on what that search term is, you are not guaranteed fisrt place in the listing. If the search term is too generic, there will be so many sites (including 'biggies') competing on it (perhaps without trying) that one is not placed well. It needs to be a bit unusual. Not something likely to be a major find in nytimes, bbc etc.

Hi Ian,

that's not my experience - but i wll look into it.

Highlighting in resulys pages suggest that engines do understand underscaore as much as hyphen or even just conjoined words. It seems to me it treats all 3 equally.

But I can check further. Thanks, because that MIGHT be an important pointer.

Steve

Hi Ian,

Offhand, it seems you are back to front here. An underscore delimits words whereas a hyphen joins them : for eample, bit-wise etc. Modern grammar is tending towards omitting hyphens and joing the two word bits into one long word. These days we dont write mis-placed but misplaced etc.

Steve

Ahh - the highlighting issue. If you check around (do a few searches on the subject), you'll find that the highlighting that is applied after a search is presented is NOT the same as the how the search engine parses things before it's presented.

It's been tested and shown many times (that's why I recommend you do a search and see the results for yourself). The reason it's been tested so often is because of exactly the assumption you made - which everyone makes. No one believes it until they see for themselves.

This is a common misunderstanding, and is one of the wierdnesses of Google. You'd think it would be the same, but it's actually two completely different functions. :(

Ian

Offhand, it seems you are back to front here. An underscore delimits words whereas a hyphen joins them : for eample, bit-wise etc. Modern grammar is tending towards omitting hyphens and joing the two word bits into one long word. These days we dont write mis-placed but misplaced etc.

You are right - gramatically.

But these are programmers, not english majors. Here is the reason from Matt Cutts himself:

http://www.mattcutts.com/blog/dashes-vs-underscores/

Ian

Caveat 2 : your new domain with name including great search eterm keywords should be a primary hosted site and not a simple redirection to a hosted site (if that makes sense) or the potential advantage will be lost.

Hi Ian,

I *seem* to be ok using the underscore, but with any of my pages stuck around position 11, and 12 it could make all the difference - i might rry changing one page to see, but it will take many weeks to see te resulys, and would require that i change all the links I put out with the old name :(

If it could get me onto page 1 from page two i would be worth the effort : )

Thans for the dialog !

Steve

No problem, Steve :)

As you can see, sometimes nice, clear logic has little to do with how these darn engines work...

Ian

Hi Ian,

yes I was alway aware that highlighting itself may mean nothing. But now i am confident that filenames do count. And underscore seems to work ok in this regard. For a domain name, just joining terms is both most common and does work - search engines can distinguish search words in a domain name formed by just joing words into one big word. Later I will relaunch under a new domain name in just this way.

Steve

And underscore seems to work ok in this regard. For a domain name, just joining terms is both most common and does work - search engines can distinguish search words in a domain name formed by just joing words into one big word.

Neither of those two statements are true, but feel free to test them - just make sure that you don't do it on something that may affect your income :eek:

Ian

Hi Ian,

yes search engines are pretty hardto fathom !

I would like to rank well for 'beijing guide'. Surprisingly, on first page is a result titled 'Shanghai Guide' and even the text text selected as a description does not show a 'beijing'. But I know that that site is by same company as one that makes a 'beijing guide' - so even a link seems enough to get a good position ! Or perhaps common server.... Or ...

Here's an out of the blue question that has been on my mind : does having images on a sire help ranking at all (after all, a 'picture tells a thousand words'). Most sites avoid images because of bandwidth. Good photos or diagrams can be valuable. Do they count for nothing in search engine algos, as seems to be the case ? Anyone know ?

Steve

For images, two things work:

1: The ALT attribute does contribute to rankings
2. The filename of the image also helps (ie beijing.jpg)

The content of the image itself does not matter, anymore than the content of Flash movies matter.

Ian

Hi Ian,

No I think you're wrong here.

I know a site that is not normally on first page (or even second) but search for the trhree words that conjoined make thier domain name (not unusual terms to search for) and they return as 1 and 2.

As I explained earlier, an underscored filename with keywords does better than same site named without undersored terms.

Of these two things i can be sure. Perhaps a hyphen would do better, but no good reason to think so right now.

Steve

Hi Ian,

With some trepidation I did add some ALTs - seems only fair given the trouble of providing an image - but not renamed images with keywords. Interesting - thanks. But to avoid overload of keywords, i better not change this now !

Steve

If you are not convinced by conjoination be a good domain name method try searching for 'beijing travel tips'. Number one is not my site. It shows my point. For other likely search terms, that site does not not appear on page one.

Search for 'Beijing guide' and number 1 site is - yes, you guessed it ! (also not mine)

POint made i think.

One example isn't very good evidence. Do 15 more searches, on a variety of things, and see what happens.

Well, I gave 2 now regarding domain name.

I have to admit that my domain name is not of this type, but i will relaunch later !

If you are not convinced by conjoination be a good domain name method try searching for 'beijing travel tips'. Number one is not my site. It shows my point. For other likely search terms, that site does not not appear on page one.

Nope, not convinced, sorry. The title tag clearly shows the term with spaces.

More to the point, I think you are overlooking this:

1. Do a search for this: allinurl: beijing travel tips - this site will not show up, yet the words are clearly in the url. Reason: Google doesn't recognize it.

2. Now do a search for this: allintitle: beijing travel tips - you will see that this site uses those keywords in it's title more than anyone else.

3. Now use this tool: http://www.webconfs.com/anchor-text-analysis.php and type in that domain, and look at all the backlinks to the site with "beijing travel tips" as anchor text.

IMO, the IBL anchor text + title text is what is causing this site to rank as it does, not the domain name. The above three checks demonstrate this.

Ian

Of course, if not covering a niche, domain name will be less useful, but will still still make a difference.

Hi Ian,

I feel this is a thrashing out of ideas.

I don't give great credence to the allinurl test, but overall maybe you have a point, maybe not.

I just can't say for sure right now. But my bones tell me that domain is critical. If i change my mind later i will be sure to tell you. I have changed my mind often enough so far ! Just want to get at the truth. Don't care a jot for 'being right'.

Steve

Doing the above analysis for 'Beijing guide' results in the same conclusion - anchor text and titles.

Typically, when you name a site mywidgetsite.com, you will link to it using "My Widget Site" as the anchor text, and it is this anchor text, which contains the spaces, that the search engine is ranking the site for.

More information on this effect: http://en.wikipedia.org/wiki/Googlebomb

No worries on "who is right" this is a forum and every opinion deserves to be heard. I've had my mind changed several times myself. I don't think it will happen on this issue, but I'm always open to new data and new views :)

Ian

Hi Ian,

Got to sleep now. Final thoughts for today is : you may be right. Sometimes all the vidence doesn;t point to the obvious conclusion - somebody's razor - Occam ?. Maybe the simplest explanation is yours. Another day ....

All the best,

Steve

Hmmm,

Ok, i can agree with your reasoning.

But that still means that choosing a domain name based on good search terms key words is a good idea !

Good night :)

Steve

IMO you're probably both right, G has always tried to draw the maximum info from an Url, difference now is they're no longer including this in their allinanchor calculations, which now makes it more difficult for us to know how/where the value lies ;-)

Someone posted the following list on another forum just now, and I had to share, since it's not only funny, but is clearly related to this thread.

You remember me saying that Google doesn't attempt to parse keyphrases in domain names and rank sites for them without dashes?

Well, the following companies are probably very happy it doesn't:

1. A site called ‘Who Represents’ where you can find the name of the agent
that represents a celebrity. Their domain name… wait for it… is
www.whorepresents.com

2. Experts Exchange, a knowledge base where programmers can exchange
advice and views at
www.expertsexchange.com

3. Looking for a pen? Look no further than Pen Island at
www.penisland.net

4. Need a therapist? Try Therapist Finder at
www.therapistfinder.com

5. Then of course, there’s the Italian Power Generator company…
www.powergenitalia.com

6. And now, we have the Mole Station Native Nursery, based in New South
Wales:
www.molestationnursery.com

7. If you’re looking for computer software, there’s always
www.ipanywhere.com

8. Welcome to the First Cumming Methodist Church. Their website is
www.cummingfirst.com

9. Then, of course, there’s these brainless art designers, and their
whacky website:
www.speedofart.com

10. Want to holiday in Lake Tahoe? Try their brochure website at
www.gotahoe.com
:D Ian

Optimizing Information Architecture is something many sites are not very good at. Most of the re-design projects I've worked on didn't need much more than implementation of the keyword research into the IA. They were optimized for the big 3 ie:Title, copy, IBL's but didn't think to optimize the foundation of the site.

Initially I added it to my optimization strategy for "on page" factors, however, then along came Google and IBL URL analysis. Then it was about more control over what gets analyzed for IBL's. I wasn't confident that IBL's would use keywords in the link text so... with optimized IA it was less of an issue.

Too often you find sites that have all or most of the files in the root of a site. Agencies are notorious for this BS, IMO, a maintenance nightmare. Why, could be because too many people read outdated information and in the old days depth of crawl was limited by folder structure. So... some smart guy started the *myth* that all files should be in root. Didn't believe it than and regardless of SEO even if it was a factor I'd still use folders because everything in root results in an unscalable site or at the very least a "dogs breakfast" of files in the root!

Early on I came to the conclusion that keywords in IA carried as much weight in the ranking algo as domain names. I never subscribed to the theory that keyword domain names were important other than the domain name is often in the link text of an IBL, so... more control over IBL link text. One factor in that decision was that those touting it were often domain registars or revenue was derived from the statement. I always try to figure out motivation for articles. If the author seems to derive revenue from a statement I don't pay much attention to it. It's biased!

Recently SE "trust" of content has been on my radar and once again I'm trying to figure out if IA optimization is a factor. IMO, SE "trust" seems to be about a lot more than just IBL's and age of sight. These two are pretty much provable, but... there seems to be other factors and IA and network/host may be among them. IMO, if a filename has the query term in it logical naming conventions would result in a file name that indicates the topic of the content and therefore relevance to the query. The question is, does SE "trust" analysis go down to the page/content level or is it just the site being analyzed. If it goes down to the page level then, IMO, Google, is quite likely including IA in the analysis as IA seems to be part of what is analyzed in many of their algos.

mcanerin - that was the funniest thing I've read in ages :D

and in regards to another post, I was of the opinion that Google at least was indexing.swf files .. and apparently

'Google can extract much—but not all—the content in the published SWF file. This content includes all the text and all the links in the file, but not necessarily in the order and in the same way you originally authored the file.'
http://www.informit.com/articles/article.asp?p=454163&seqNum=1&rl=1

- I have yet to see them indexing much other than URL's / links in the flash...and I've been poking around

....they are listed similar to PDF files but with [FLASH] as opposed to [PDF] prior to the URL link, but I am still unconvinced that Google CAN'T read more in the Flash. Do you have any more info? And do you think this is likely to change? or is xml functionality going to nullify the need?

cheeky first post - sorry. Been an avid reader for ages, and I guess the giggle got me motivated to get involved :o

Hi MacCallow, and welcome!

Regarding Flash files, here is a fictionalized but essentially accurate conversation about Flash being indexed by the search engines:

-----------------------------

{Meeting Room at SES}

Flash Developer: I understand that Macromedia has released a set of programs that will allow Flash to be accessible to search engines.

Search Engineer: Yes, that's true, we've looked at it.

Flash Developer: So the search engines can index Flash, right?

Search Engineer: Technically, yes, even without the kit, in the same way we can index PDF and other non-HTML documents, but...

Flash Developer: Yippee! My job is saved!

{Flash Developer runs out of room to tell all his friends and blog about it}

Search Engineer: But... As I was saying, the vast majority of Flash content is not indexable and takes up a huge amount of bandwidth for very little gain in the relevancy of the SERPS, therefore although we CAN index it, we DON'T index it...Hey, where'd he go?

Audience Member: He went to go generate a bunch of work for SEO's... ;)

-----------------------

The bottom line is, that if you think about how the typical Flash file is set up, the first bunch of frames usually just say "Loading" and "1%" 2%" etc. Then often there is some sound files and graphics loaded during this period, followed finally by content, which is often a "symbol" of text buried in a separate button script, rather than editable text. This is usually streamed.

In short, a search engine may have to use up several megabytes of file just to get to the part of the Flash movie that contains text, only to find out it's a program, script or vectorized text, rather than what it would consider to be content.

This is hugely inefficient, a waste of bandwidth, and apparently several studies by the search engines have shown that it's simply not worth it.

Technically, a search engine could also attempt to run scripts on webpages, or download images, vectorize them and then run OCR software on the images to see if there is content there, or maybe run sound-to-speech detection of video and sound files, but the amount of processing power to accomplish this is far in excess of the benefits they would likely gain from it.

Finally, due to the structure of a Flash file, it's difficult to know the context of any text you may find there, since the text may be saved as variables or just pulled from the server as necessary (since a search engine doesn't execute Flash anymore than it executes Javascript, that text would never get pulled from the database or flat-file and thus would never be in the Flash in the first place).

Then you have the issue of Flash files executing other Flash files, and so forth. Since a search engine doesn't execute programs or scripts for safety reasons, none of that content would get indexed.

So yes, it's possible. But it's not gonna happen, not for a VERY long time, if ever. Maybe when bandwidth and processing power becomes infinite and free, or close to it.

As for finding URL's, etc - There are two ways that can happen. First, when you generate a Flash file, Flash creates an HTML "container" that calls it. You will notice that it often has content pulled from the Flash file in there as a comment, or noscript. Second, I'm told that if the URLs are in the first couple of frames, a search engine may choose to peek in there, but I've never actually seen that happen personally.

I certainly would not count of either method to rank my website in a competitive area, however. Noscript would work better.

There are some workable methods that you can use to present the search engine with the content from a Flash movie, but hoping it will do it on it's own is not one, IMO.

Ian

wow... that was a way more involved and 'useable' reply than I had anticipated! Thanks mcanerin :D

Can I push you on whether you think having both html and Flash, or just making use of XML functionality is better IYO?

I should probably admit at this point that I have a basic understanding of HTML, a pretty good understanding of how on page things like content, meta data, titles etc work, realise the importance of relevant backlinks, RSS feeds and blogs and so forth... and am quite clued up on reading about the development/structure issues - about 10% of which I actually understand, but I know what should be done without knowing how to do it... frustrating yes, but only been at it for 14 months... I'm getting there. I'm sure I've left out a few things that are terribly important, but its Friday... I have a glass of wine... the kids are finally asleep... and I'm functioning on minimal brain capacity... not all that unusual... I'm actually still waiting for my IQ to leap back 20 points after the twins were born... but that was 3 years ago and I'm slowly and reluctantly realising that this - wel, mite knot hapen....

As such your reply was pure nectar. I actually understood it! SO important these days.

.. oh, andd tanks fore the welkom ;)


oh.....and last point - without taking up another post - Danny Sullivan - you are like one of my unwitting mentors - I wanted to say how sorry I am that you're leaving SEW (from Matt Cutts blog - shoot the messenger eh... no - happy for the heads up) , but that I really hope that your next endeavour is rewarding, prosperous and - well just right for you. I'll miss having you around here... daft from a new postie, but like I said - I've been a silent witness for ages.. Go for it! and God bless

Place seems suddenly inundated with Canadian-Scots ;-)