Anchor text penalties

[Tonearm]

I remember reading that Google will penalize a site if it has one of the following:

a) different anchor text pointing to the same page

b) the same anchor text pointing to different pages

I can't remember which it was, or was it both?

[tedster]

Assuming you're talking about internal anchor text, b) is a lot more likely than a) - but neither one is a sure thing at all. b) just sends a confusing signal which might cause some ranking loss for one of the pages (but that's not a true penalty.)

Different anchor text pointing to the same page is pretty common and can be innocent - but it could also be the kind of manipulative SEO that Penguin doesn't like if it's over the top.

[indyank]

Different anchor text pointing to the same page is pretty common and can be innocent - but it could also be the kind of manipulative SEO that Penguin doesn't like if it's over the top.

lol...I always felt that way. People were talking about google considering a lot of external links with same anchor text to a page as being over SEOed but I would say that it is more natural than having external links with several different keyword variations as anchor text. Third party genuine linkers would in most cases link to a site with some relevant keyword(s) present in the h1 or title of the page being linked to, than inventing their own keyword variations to link to them. It is thus more natural than getting linked with a large variation of similar/related keywords as anchor text. But who knows what/how the crazy over thinking Google PHDs think about over optimization!

[louieramos]

Just keep the balance of your keywords and backlinks and you won't have any problem. I know this by experience on my sites and my client sites. Also make sure you complement this by great site structure and contents

[indyank]

But doesn't having a balance suggest a "managed show"?

1) Links with title or h1 or keywords in title or h1 as anchor text.
2) Non-keyword anchor text eg; More, This, Buy, download, etc.

Doesn't the above two sound better?

[neildt]

So are you saying too much variation in external and/or internal could cause a penalty ?

I thought Google pengiun update was to do with over optmised link text and anchor text that was not diverse enough eg 20 occurances of the same anchor text.

[Robert Charlton]

As I remember, the Historical Data Patent talks about anchor text that is not diverse enough, or anchor text that is too diverse.

[indyank]

So are you saying too much variation in external and/or internal could cause a penalty ?

IMO, anchor text for internal links is not being scored currently. Internal links just count for flow of PR but not flow of anchor text.This I think is one big change they did around Panda time last year. Either this could be part of panda or a separate change. But big changes related to links were mentioned by a google employee on a popular forum and my guess is this is one of those changes then.

So it all boils down to anchor text flow via external links. Google does use other text on the linking page and not just the anchor text for determining this flow. However, with penguin, they seem to have introduced an algorithmic penalty based on anchor text used. It might be based on over use of same keywords or diversity in keywords used as anchor text.But my argument is that overuse of same keywords (from multiple domains) is a natural one as long as the keywords are present in title or h1.But google engineers might not necessarily be thinking like I do.

[BeeDeeDubbleU]

Different anchor text pointing to the same page is pretty common and can be innocent - but it could also be the kind of manipulative SEO that Penguin doesn't like if it's over the top.
&
As I remember, the Historical Data Patent talks about anchor text that is not diverse enough, or anchor text that is too diverse.

Most of my sites are generally small, straightforward business type websites with a limited number of pages. I usually have a main menu and a footer menu with text links like "About Blue Widgets". These would appear on every page.

Google surely cannot penalise sites for doing this when it is standard practice on millions of websites?

[indyank]

IMO, internal links are not treated the same as external links for flow of anchor text.

[Tonearm]

I'm changing some of my links from this:

Blue Widgets
Red Widgets
Yellow Widgets

to this:

Widgets
Blue
Red
Yellow

This means I'll also have:

Donkey Saddles
Blue
Red
Yellow

So I'll have anchor text "Blue" pointing to my Blue Widgets page, my Blue Donkey Saddles page, and many others. Am I asking for trouble?

[theskunk]

Can anyone confirm excatly what factors you use to determine is a link is good or bad. How do you determine if the site linking to you is good or bad. Is anchor text saying 'buy blue widgets' worse than anchor text saying 'mywebsite.com'.

Google made this mess with their drive to demonstrate the link vote, and now we are all running around like headless chickens cleaning up their mess. It's one big joke. It like divide and rule... only google doesn't divide they just jumble us all up by giving us a shake.

Should I Spend the next 4 weeks of my life trying to resolve 15 years of optimised links? Or should I ASSUME that google is clever enough to just ignore them ... ?

[edited by: tedster at 2:12 am (utc) on Jun 13, 2012]

[Squilookal]

The key here is anchor text diversity because with naturally acquired links there is always diversity as they come from different sources and are written by different people. Google won’t penalise for mywebsite.com anchor or non keyword specific anchor text like “here” or “more” no matter how much you have but you would be more likely to fall foul of the algo if you have 90% of your backlinks with a single keyword or phrase like “buy blue widgets” for the simple reason that organically acquired links are never like this.

A very good article I read (with a lot of actual data rather than just opinions of self proclaimed SEO gurus) suggests that having not more than 50% of your anchor text as your target keyword is safe and that the SEO drive should now be less related to simple anchor text and more concentrated on niche and content relevancy. To me this makes sense as it is how natural links get built and it is more difficult to replicate. For those that use SEO tools to look at back link profiles, it is nearly always clear if links have been bought or SEO has been done. What we’re seeing is just Google’s algo finally catching up and doing with millions of metrics and calculations what the human eye can see in a couple of seconds.

Assuming Google is clever enough to ignore “bad” links was once what we all relied on of course but negative SEO is here, what isn’t clear (to me at least) is how Google treats older “bad” links that were placed before the algo updates. My experience is old bad (over optimised) links are largely ignored but new bad links cause a slap. In reality of course, its probably much more complicated than that because it always is.

[deadsea]

Internal anchor text diversity does not hurt me currently. I few years ago, Google was often ranking the wrong page for queries. The search terms would all be on the page that was ranking, but there would be a better page much more suited for the query. At the time, I fixed the problem by interlinking all my pages with anchor text of all the search phrases that Google was pointing to the wrong pages. pointing to the page I would prefer to rank for that phrase. At the time, it worked like a charm. Google looked at the internal anchor text coming into the page, and payed a lot of attention to it. Some of my pages have as many as 30 different related phrases linking to them internally.

I'm not sure this is helping at all today, but the site wasn't penguinized or pandalized either.

[gouri]

Some of my pages have as many as 30 different related phrases linking to them internally.

There is another possibility with internal links, and I have seen both options considered the good one so I would appreciate your thoughts.

If a page on a website is linked to from the body text of two pages, is it better to use the same anchor text for both links or different anchor text?

Example: If Page 10 on a website is linked to from the body text of Page 3 and Page 6, is it better to use build great abs as your anchor text twice or is it better to use build great abs once and define your abdominals once?

Could using the same anchor text twice be considered over-optimization?

[realmaverick]

Ok I've got a question about breadcrumbs.

Widgets > Brand Widgets > Brand Model Wdigets

OR

Widgets > Brand > Model

I am guessing the second option is less spammy, but also less context?

[realmaverick]

Actually, scrap my question. I think it's pretty obvious that Widgets > Brand > Model would be a safer bet.

I've never really thought about it, as they're automatically generated. At present they are the former example, I'm going to switch that around.

[indyank]

I'm changing some of my links from this:

I'm not sure this is helping at all today, but the site wasn't penguinized or pandalized either.

Whatever that you do with internal links might not help these days as I believe that google recognized internal links for only PR flows and not for anchor text.

[indyank]

Google won’t penalise for mywebsite.com anchor or non keyword specific anchor text like “here” or “more” no matter how much you have

Agreed.

The key here is anchor text diversity because with naturally acquired links there is always diversity as they come from different sources and are written by different people.

I am not sure why you feel so. Despite different people writing them, most often they are influenced by the keywords in the title or h1 of the page they link to. This is more natural.

Do you meant different combinations of keywords in title or h1 and verbs as diversity. For example if "widget" is the keyword in title or h1, then third parties linking to your page as "buy widget", "get widget", "try this widget" etc. are fine and looks natural. But if those diverse sources are linking to the page with keywords similar or related to "widgets", then id doesn't sound natural as the above examples.

[realmaverick]

I am not sure why you feel so. Despite different people writing them, most often they are influenced by the keywords in the title or h1 of the page they link to. This is more natural.

I disagree. If you look at a reasonably large site, that hasn't engaged in any link building campaigns, you'll nearly see "Brand" being the number 1 anchor text. You'll also find "brand.com" ranking pretty highly on top anchor texts too.

Some kind souls will of course link to you with the target keyword, but these will be few and far between, which is what you must mimic, when you're link building.

If even 50% linked with the keyword, it's too much IMO. I've always felt this way, long before Penguin.

[indyank]

realmaverick, i have already agreed with what you are saying. What I am talking about is the other ways in which you are linked with anchor text and which looks more natural.

[Squilookal]

In your example indyank one could consider editorials that link using some sort of call to action like “buy widget” or “get widget” as great but only in moderation. These days I’d probably be happier with a higher percentage of links that are something like “there are some great widgets around at the moment, this site has a good selection” with “this site” being the anchor and hoping G can interpret the text wrapping it and the other elements on the page as providing the context.

Personally I think that anything that either is, or closely mimics natural/wider anchor text diversity is now much more important. Writers giving the genuine editorial links Google wants to use as a basis for ranking sites have a lot of ways of linking at their disposal and it is my belief that Google’s algo now puts greater weight in the overall context of the site and page the link is in rather than the anchor text. I think they have started to do this now because they’ve got better at the bigger picture interpretation than was once the case and anchor text was the “easy” and less resource intensive way of doing it.

To get the general weight of relevance concentrated more directly to our target keywords I still think keyword anchors are also necessary to guide Google but in a lower percentage than we would have aimed for in the past. I will probably be aiming for the 50% mark as an absolute max as I feel this is still below penalty levels but as realmaverick says, this is still unnaturally high. The sites that are doing best in the keyword space I monitor have a very low actual keyword anchor text proportion and most links have random anchors that come from highly relevant pages. In most cases these are still all paid links or from in house networks but they are just not hammering keyword anchor text so aggressively.

[1script]

Regarding what's natural in links: I think THE most natural is not even the domain name or "here" and obviously not title or H1, sorry indyank, it sounds like a bit of wishful thinking here that "normal" people would even look at the title or find the H1 which can be anywhere on the page.

Vast majority of my links are simply URLs of the pages. It's rather logical, too: in order to create ANY link with anchor, even with such simple one as "here", one needs to go through the trouble of either editing the HTML code or at the very least feeding two things into some kind of a tool that converts text into links (think Wordpress post html editor or forum post editor). There's nothing simpler than just copying the URL from the address bar and then pasting it into your post text, assuming that in the final version of the post it will be automatically converted into a link.

I'd say unless you are dealing with HTML-savvy crowd, vast majority of your naturally accrued links will be simply URLs - the URL for the href tag and the same or shortened version for the anchor.

That presents its own problem here: I, for example, since the dawn of time have been using the same phrase for the title, H1 and the URL, URL-encoded, obviously and spaces converted into dashes. I think I'm being penalized for that now because it may now appear that I am over-optimizing for the phrase. I have no idea what to do now because de-optimizing will mean changing the URL structure of the site and that sounds like a death penalty for the site. What I mean by that is that, given the fixed crawling budget, Gbot will be busy ONLY learning about URL changes and following redirects for months to come. At the very least, the immediate consequence of a site-wide URL change will be the cut of effective indexing rate in half.

Anyhow, does anyone else here see a potential benefit of such drastic de-optimization as removing the title phrase from the URL and thus changing the whole URL structure of the site?

[realmaverick]

Yup URL and brand will nearly always appear first. Though in my experience, it's not [domain.com,...] it's more domain.com.

[lucy24]

:: trying to make sense of this without proceeding directly to "wtf?" state ::

Same words linking to different page:

"more information" (about subject I've spent the last two paragraphs discussing) >> link to fragment on Page A
"more information" (ditto, with different subject) >> fragment on Page B

vs.

"more information about red widgets" (which I've spent the last two paragraphs discussing) >> fragment on Page A
"more information about blue widgets" (ditto) >> fragment on Page B

Different words linking to same page:

"red widgets" (which are related to but not directly mentioned in current discussion) >> link to Page A (whole page or fragment)
"blue widgets" (arising out of different discussion) >> ditto

vs.

"red and blue widgets" (where current discussion gives rise to question of red widgets only) >> Page A
"red and blue widgets" (ditto, but blue widgets only) >> ditto

Now can I get to the "wtf?" part? The link text says what the link text needs to say. How 'bout if we go retro and change them all to either "click here" or {exact title of page}?

[realmaverick]

Lucy, while you're right. And I personally feel we should be able to link to our content with WTF we like, without EVER having to consider Google, unfortunately we can't.

[theskunk]

@ Squilookal - thanks for your incitefull response

This is quite a nice thread - and shows there are a lot of ways to think about the same question, but not any answers (to be expected at this point). I wonder if anyone can reccomend the best process post YAHOO explorer for getting a list of your back links and a list of the anchor text. Is there afree tool or combination of tools? Since once you get the CSV together you can sort by ANCHOR text and get a clear picture of your ratio's and start the fix.

Whilst google might ignore internal anchors I wonder how they deal with links from sub domains to root and vice versa that are from same owner.

My whol architecture uses sub domains for shop and support etc and they interlink deeply with 15 years of optimised anchor text - I always used keywords for links - long before google.

After 4 weeks I just cant get my head round the best way to address this without making matters much worse for my ECOM company that is now running at 45% revenue of last years. Employing 5.... soon to be 1. (i'm not hear to moan). But I wld like to find some new best practise guidlines and determine if there is ANY merit of fixing legacy links to the new IDEALS.

[lucy24]

Concrete question: Does it make any difference whether you're talking about menus (or functional equivalent), as opposed to, ahem, natural links within page content? I can definitely see where having different menu labels leading to the identical page could look like you're up to some kind of hanky-panky. Unless of course your name is amazon or similar.

[deadsea]

There are (non-SEO) reasons to vary anchor text:

1) Context: "Blue Widgets" is great anchor text, unless it is in list of Widgets, in which case you probably want to say "Blue" and forget the redundant "Widgets"

2) Alphabetical indexing: In your alphabetical list of products you might want to have "Blue Widgets" and "Widgets (Blue)" so that users can find your product either way.

3) Extra descriptiveness: Sometimes you might have to say what it is for: "Blue widget for Cleaning"

4) Multiple Taxonomy: You may have blue widgets listed in multiple categories. If this is the case you might need to describe them differently "Blue widgets for cleaning your car" vs "Blue widgets for cleaning your kitchen sink". In some cases products are just called slightly different things depending upon use.

5) Synonyms: If there is more than one name for something, you might have to link to it different ways, even in the same menu.

6) Different editors: You have multiple editors working on the site that are free to make editorial decisions.

7) User generated content: users will certainly link to stuff differently if you let them create content internally. One site that I worked with even had user created lists of products and users often edited product names in those lists for consistency or brevity.

[Oimachi2]

I have an exact domain match for a non competitive industry.

I was ranking #3 so figured I'd do an article markeing run (only about 10 of them, the large ones)

I used "natural" links, no money keywords. To my horror Google dropped me 14 spots immediately...

Seems they don't like ANY kind of link building.

Seems like diluting exessive anchor text will sink you further.

Seems like getting out of SEO completely is the solution.

PS. MSN Adcenter converts wayyyyyyyyyyyyyyyyyyyyyyyyyyyyy more than adwords.

Adwords is a waste of time, Google is a waste of time.