homepage Welcome to WebmasterWorld Guest from
register, free tools, login, search, pro membership, help, library, announcements, recent posts, open posts,
Become a Pro Member
Visit PubCon.com
Home / Forums Index / Google / Google SEO News and Discussion
Forum Library, Charter, Moderators: Robert Charlton & aakk9999 & brotherhood of lan & goodroi

Google SEO News and Discussion Forum

This 56 message thread spans 2 pages: < < 56 ( 1 [2]     
The Power of Text Navigation - a case study

 8:15 pm on Sep 18, 2006 (gmt 0)

Most people in this game know that anchor text is more powerful than image links, even image links with a good alt attribute. But just how much more powerful is it? I recently had a chance to test this in a relatively controlled situation.

The site was already successful in their niche - and it's pretty competitive, with 400 million Google results on the big trophy keyword. The pages looked sharp, but the main navigation (top and left) was created with images of text with alt attributes, not anchor text. And javascript rollovers, naturally.

The only change so far was to switch this setup to CSS styled text menus (links in list elements) a few weeks back. And funny thing, the look is so close to the original that no end user or affiliate has yet commented on the change. Some of the business staff haven't even noticed.

At any rate, with the Google changes on Sep 15, we got our first look at the new rankings. On the trophy keyword, the home page went up about 45 positions, from out of the running to a second page contender. With lots more improvements to go, this picture is quite hopeful. The power of anchor text on the main navigation turned out to be BIG.

Lots of other major keywords went to #1 or into the top 3, from previously being 10, 20, or more positions down. In all, it looks like 200 phrases have improved from the benchmark sample we are tracking. I'm sure as the new conversions start to show up, we'll learn about lots more keywords that are worth tracking and tweaking, too.

Lesson for me -- don't let the issue slide when it comes to using text in main navigvation. It's a lot more important than I used to think, and I used to think it was pretty important anyway.

[edited by: tedster at 1:38 am (utc) on Sep. 19, 2006]



 2:43 pm on Sep 19, 2006 (gmt 0)

I totally gave up using images for links a little while ago, figuring it may or may not help in the serps but it definitely wouldn't hurt. Plus I think it looks better; I like my navigation to be clear and crisp, no fuss.

Now I wish I could go back and redo all the sites I've ever done with everything I've either figured out or learned here.


 3:02 pm on Sep 19, 2006 (gmt 0)

Tedster is talking about repeating KW in the anchor text, BigDave about repeating the anchor text itself.

this is bad:
<a>keyword keyword keyword widget keyword</a>

this is OK:
<a>keyword widget</a>
<a>keyword foo</a>
<a>keyword baddabing</a>
<a>keyword kapow</a>

Did I get that right?

Where in the spectrum of good anchortexting does this fall?

¦- <a>widget</a>
¦- <a>foo</a>
¦- <a>baddabing</a>
¦- <a>kapow</a>

... assuming all the proper nouns make sense when arranged like that.

This menu is not passing any "keyword" anchor power to the baddabing page. but it is passing concentrated baddabingness. Sometimes that's enough :)


 3:17 pm on Sep 19, 2006 (gmt 0)

I think this study could go beyond the effect of text navigation on search rankings to include click-through rates. I have noticed a decline in CTR when I use images that contain a call-to-action compared to a text link containing the same phrase, positioned in the same spot where the image originally was.


 3:36 pm on Sep 19, 2006 (gmt 0)

Ah, Tedster, why'd you have to go and do that. I quite liked having people chase recips while I focused some attention on internal anchor text. It has been working for me for at least three years now.

But, since we are spoiling the day, don't forget the power of in-line links back to the site's own relative pages. It is not just nav menus that carry this effect.



 3:40 pm on Sep 19, 2006 (gmt 0)

*Did I get that right?*

Your first example where the KW is repeated throughout the navigation links would be the iffy one, IMO.


 3:45 pm on Sep 19, 2006 (gmt 0)

webwitchy, I am starting to think that their is diminsihing return in repeating the keyword too frequently in the nav menu...

widget history
widget faq's
widget suppliers
widget sales

...may well be a bit too much. I am watching a site slide (not a lot, but noticably) with a menu much like that. It has been top three on the single word terms for several years. Right now it is #5 for the plural term and #11 for the singular.

The singular term shows in 4 of 8 top level menu items. The plural in 3 of 8.

I strongly suspect that this is related to the slip, but it still rates well enough that I am afraid to tinker.



 4:11 pm on Sep 19, 2006 (gmt 0)

- <a>widget</a>
- <a>foo</a>
- <a>baddabing</a>
- <a>kapow</a>

This would be my preferred one, if it was a much longer list of more than 3 or 4 of "keyword". You will, of course be using "keyword" in your title, and you are also making an upper level page all about "keyword" that will then have some inline text links down to "keyword widget".

You don't have to shove everything down the SE's throat. Even MSN is starting to figure some of this stuff out, and google has been there for years.


 4:37 pm on Sep 19, 2006 (gmt 0)

For a search engine, there is a difference in trustworthiness between text links and alt-attributes of images. Alt-attributes can be manipulated, as they're not usually visible. They're like Keywords metatags in that respect. Text links are WYSIWYG.

So they value text links more than alt-attributes.


 5:07 pm on Sep 19, 2006 (gmt 0)

Unfortunately, using techniques like this may also cause your site to get zapped in google. This has been the case for a number of sites in yahoo (which seems to have some sort of filter for this) and I've already seen it happen to a few sites in google. The main problem seems to be overuse of the same keywords in anchor text as someone else in this thread has already warned against. I'd be very careful about changing my site's navigation, particularly if there are thousands of pages to the site. Google may perceive this as massive link manipulation and lower the boom.


 5:27 pm on Sep 19, 2006 (gmt 0)

This would be my preferred one...



 6:47 pm on Sep 19, 2006 (gmt 0)

This is a tricky one, and the line between keyword stuffing and what otherwise seems like a logical navigation structure is completely blurred.

Consider my example. I have a site which promotes widgets of over a dozen different colors in hundreds of cities. My nav structure for the last 5 years has been similar to pageoneresults' above, just without the heading tags and list. Everyone else and their momma who comes up for searches like "widgets in miami" has tons of links repeating the keyword "widget" for lists of cities of varying sizes.

ALthough I considered this a total Google landmine, just recently it occured to me:

If my main Kansas widget page links to 100 pages of widgets in cities throughout Kansas, then it is almost stupid of me to link to them only using the name of the city. To support this further, now take the whole structure down one level to "blue widgets in miami". If I have 9 different color widgets in miami, should I not link to each with the color and city in the text link.

Logic says yes. Let's see what Google says.


 8:59 pm on Sep 19, 2006 (gmt 0)

Here's my guess -- you'll do great for [blue widgets in Kansas] or even [Kansas widgets] but if you were ranking well for [widgets] alone, that will take a nose dive.


 2:30 pm on Sep 20, 2006 (gmt 0)

I've found that if you while repeating the keyword in a menu setting may not give you a boost, it almost always gives you (well, me anyway) an indented listing in Google if the destination page has a similar title tag as the anchor text.

Again, this worked for me, may not work for you.


 4:56 pm on Sep 20, 2006 (gmt 0)

I think the line between keyword stuffing and a logical navigation structure is not necessarily blurred. Rather, it is a question of how smart the Search Engines are -- or how easily they can be tricked.

My impression is that SEO techniques that worked well in the past tend to stop working, or begin to hurt, as the SE's get smarter.

Consider the earlier example, it was a site which promotes widgets of over a dozen different colors in hundreds of cities.

The nav structure for the last 5 years has been similar to that of its competitors -- lots of usage of both "widgets" and "city name" in the navigation, making it easier to come up for searches like "widgets in miami" and perhaps even for the keyword "widget" (since this keyword was being used in the anchor text for nearly all of the navigation links on the site.

This nav structure may have worked well for SEO purposes, but it really isn't the most logical, or ideal nav structure for users. Absent any SEO concerns, it really would be more logical for all of these internal links to only use the name of the city -- assuming the site is devoted to widgets of various types and locations. In other words, the user already knows the context is one in which the focus is on widgets, so there isn't a good reason to constantly mention the word widgets -- particularly in the navigation, where it is redundant noise that reduces usability.

Thus, if the SE's get smart enough, the presence of all those redundant "widget" references in the navigation links is a dead giveaway that the site owner is trying to impress the SE's, not trying to make the site better for users.

Which, we will recall, is exactly what Google warns us not to do.

Basically, to look natural and user-oriented, the site should have the keyword "widgets" sprinkled around the site as much as necessary to form coherent sentences and paragraphs, but it shouldn't be overdone. And, it certainly doesn't need to be mentioned in nearly every internal link.

From a user perspective, a menu that reads: "blue widgets", "red widgets", "green widgets" isn't very logical if the site specializes in widgets. Of course, if this is a general site selling all different sorts of things, perhaps you might see that navigation, but more likely, there woudl be a hierarchy, with one button or tab labeled "widgets" and a subset of buttons or links below that which list the various colors.

Now, for highly obscure combinations, like "blue widgets in miami", it might make sense to use the full phrase in navigation, but more likely this will also be less than fully natural.

So, the real issue is whether Google is smart enough to detect the difference between navigation which is designed strictly for users, and navigation that is largely or entirely designed to help the search engines realize this is a widget site. It used to be important to mention "widget" everywhere you possibly could. Now it may be, or at least it will soon be, smarter to be a little more circumspect in our efforts to help the SE's realize our site is all about widgets.


 6:30 pm on Sep 20, 2006 (gmt 0)

This "keyword overuse in navigation" seemed to be one of the areas Google began working with as long ago as the Florida update. As I see it, this is not bout a "penalty" but about filtering the 1-word search results to protect them from being heavily manipulated.

Sometimes the longer phrases are more important for the business and for conversions than the single word search. I have one case where this is exactly the case. They were filtered way down on the 1-word search long ago, but those searches weren't bringing in the real money. And so we let it be, and even went after more phrases.

Also, I've had some questions about whether heavy keyword repetition in the navigation might make a page go into the Supplemental Index as duplicate content. I have never seen that, and the way I see things, that wouldn't make any sense. It's hard to give a definite yes or no to almost any Google question, but in my mind this is a definite no.


 3:01 pm on Sep 21, 2006 (gmt 0)

Kudos econman, if the site is devoted to widgets, then indeed it is logical that a "phoenix" link implies phoenix widgets. My gripe is that half of the sites that knock me into the second or third page for queries I have tracked for years have tons of these "phoenix widgets" links. I have resigned to the old, if you can't beat them, join them tactic, and started to do the same thing. I will get you yet, phoenix.wigets.phoenixwidgets.cheap-widgets.org!

Thank you tedster - it's certainly a bit comforting that you haven't come across evidence of these stuffed links causing penalty. One of my oncerns was that the keyword density for "widgets" on the page would be weighed down unnaturally, but I do not believe that hyperlinked text is treated the same way as regular text. I'd love to hear otherwise, if anyone is of a different opinion.

Oliver Henniges

 4:48 pm on Sep 21, 2006 (gmt 0)

just one question: On many occasions I use this construction:
<a href = "blue-heavy-metal-widget.html"><img src = "images/img1234.jpg" alt = "blue heavy metal widget"><br>widgets<a/>

this repetition of alt and anchor text with sometimes the anchor text being a bit shorter: Is that good, bad or neutral?


 4:58 pm on Sep 21, 2006 (gmt 0)



I see an awful lot of this myself. Geesh, I was looking for something quite specific this morning and goat back about 90,000 (no exageration) responses that all looked like city.county_name.state.zipCode.sameFrigginSite.com

The results were absolutely useless. Pathetic...

Way OT, but G really needs to get the subdomain issue under control.



 6:42 pm on Sep 21, 2006 (gmt 0)

my oncerns was that the keyword density for "widgets" on the page would be weighed down unnaturally

Oh, it can be. And it can get your page filtered for the "widgets" results. But you can still be #1 for "Phoenix widgets".

In short, a filter (which can be search specific) is not a penalty aginst the entire domain -- at least that's what I see.


 7:30 pm on Sep 21, 2006 (gmt 0)

"have you ever tried SIFR? "

I'd heard of that before I think, but it looks very cool. Anyone using it?



 7:50 pm on Sep 21, 2006 (gmt 0)

Yes, I have used both SIFR and Fahner Image Replacement (for non-Flash images). They work well and have been in place on several major websites for a couple years.

Warning: if you tke this approach, you my be tempted to "embellish" the text version of the graphical content, just to help Google along a little bit extra, you know? Google most definitely considers that hidden content and such usge will not pass a manual check.

However, I have not used them in text navigation, but only for heads and sub-heads where they can enhance the page design with with a nice display font. I suppose they could be employed in a menu/navigation area - no technical limitation I know of.


 6:23 am on Sep 22, 2006 (gmt 0)

tedster - Do you think it makes a difference on which side of the page the navigation is? Left , centre , right etc?


 3:48 pm on Sep 22, 2006 (gmt 0)

I work with sites that have both flavors of menus and I can't see a ranking difference. Right hand menus stand less chance of ending up in the site description snippet, I suppose. And there is some little sign that right nav sites may be just a bit sitckier for the end user, but nothing conclusive there.


 4:22 pm on Sep 22, 2006 (gmt 0)

We started to switch to CSS Layout w/o tables and converted our layout/navigation images to CSS backgrounds about a year ago. Also we use an unorder list <ul> for the navigation that has a class applied to it.

But my question is what type of method do you apply to remove/hide the text?

We use the first method below, and I think our reasoning was that the search engines would not register the tag as hidden or not displayed. We are not trying to blackhat anything, we are simply trying to make clean html for the SEs and visual pages for the user.

text-indent: -9999px;
[We currently us this one]

- or -
visibility: hidden;

- or -

1) Is one better then the other?
2) Have you begun to see any abuse of this CSS layout method by spammer sites?
3) Do search engines read the CSS file?
4) Would the search engines be savy enough to figure out if a site is spamming or is using the CSS hide methods honestly?

Any thoughts on this are appreciated, new territory in a way.


 6:32 pm on Sep 22, 2006 (gmt 0)

I'll take a shot at it:

1) The key is not the "hiding" method, but the fact that some mechanism (like Flash turned off) will make the text visible to some users.

2) Yes

3) Not often, so my guess is not in an automated way -- yet

4) Not by algorithm, at least not yet. But they re smrt enough to let the algo flag some urls for a hand check.


 6:57 pm on Oct 2, 2006 (gmt 0)

If you have worked in SEO long enough you will get clients who don't want to touch their pretty graphical navigation. There is an easy solution. Create a text based page footer using your keywords as anchor text to all major sub pages.

Because on page location is a relevant factor, try positioning the footer in the code at the top of the page, similar to a top navigation. You can uss CSS and absolute positioning to have the page footer hide nicely beneath the contents of your page so it doesn't really alter the look and feel of the site.

I have employed this with a couple of sites and have seen very good position increases fairly quickly.

<Sorry, no urls.
See Forum Charter [webmasterworld.com]>

[edited by: tedster at 7:15 pm (utc) on Oct. 2, 2006]

This 56 message thread spans 2 pages: < < 56 ( 1 [2]
Global Options:
 top home search open messages active posts  

Home / Forums Index / Google / Google SEO News and Discussion
rss feed

All trademarks and copyrights held by respective owners. Member comments are owned by the poster.
Home ¦ Free Tools ¦ Terms of Service ¦ Privacy Policy ¦ Report Problem ¦ About ¦ Library ¦ Newsletter
WebmasterWorld is a Developer Shed Community owned by Jim Boykin.
© Webmaster World 1996-2014 all rights reserved