How effective do you find anchor text for ranking, especially in Google? Let's cover inbound links and anchor text for link building purposes as well.

Anchor text is very important with Google, but you can over do it. Last figure I heard was no more than 50% containing keyphrase. Use semantics to vary actual text and use keyphrase in close proximity. Basically mix it up and avoid "patterns", they need to look quite random. HTH

link text is huge. :)

likely the #1 factor for many competitive phrases

There was talk back aways that incoming anchor text might cause problems after Google's Florida update. I've not seen any problems with anchor text myself, works just fine from what I've experienced.

Anchor text is very important with Google, but you can over do it. Last figure I heard was no more than 50% containing keyphrase. Use semantics to vary actual text and use keyphrase in close proximity. Basically mix it up and avoid "patterns", they need to look quite random.

I want to examine a couple implications in this statement.

First, is the method Google or another SE would use to carry out such an analysis on keyphrases in anchor text. Would the analysis only kick in once a certain number of links had been indexed? For instance, what if a site had 3 ibl all with the same anchor text, is that enough links to reach analysis stage? Would a penalty or dampening factor be applied to links or to the site/page itself? Why a 50% threshold? Would the analysis differ for one, two, etc word anchor text phrases? For instance, a seven word anchor text phrase occuring 50% of the time may be more suspicious than a one word phrase. If you had 1000 ibl's with the anchor text "www.mysite.com" or "click here" is there a justification for receiving a penalty or dampening effect?

Second, does anyone have any research supporting the fact that SE's may be analyzing keyphrase frequency percentages in ibl's and taking some action if a certain percentage is reached?

Jon

Jon

"I want to examine a couple implications in this statement."

This "statement" is like virtually all statements in SEO/M it is an opinion, but one borne out by my experience which I thought I would share with Daria.

Two or three years ago I went mad gaining ibls using my main keyphrase as anchor text and it worked very well for a couple of years. About 18 months ago in conversation with top UK SEOers (known to most here) I was advised to mix 'em up a bit. Being very highly positioned in a very competitive wordspace I was wary about changing my inbounds so I dragged my heels a bit. Around this time I noticed a slow down in success with clients' success using my then highly successful formula.

Later, when talking with a couple of WW mods at PubCon the subject cropped up again and a threshold of 75% was mentioned, their considered opinion. Still I resisted changing things too much. However the subject started to crop up on various forums and I read them all and started to take these opinions on board, after all, how unnatural are many hundreds of perfect inbounds - ie phrases that match exactly titles, H1s, etc ? ;-)

Around the Florida/Boston updates I started to vary inbounds for new clients and the results were good, but several existing clients whose main keyphrase was also the company name suffered badly, ie "The Red Widget Company".

Soon after I bombed out. I immediately got to work changing ibls (wherever I could) and eventually started to climb back. I was happy with my on-page optimisation so the improvement must have been down to ibls or Google rolling back to a less severe version of the algo. Looking back it shows my lack of common sense after being given good advise from top players, SEOers who ride out these major shifts without making big changes.

With regard to research, I know of none on this subject, but I do understand the concepts of "standing out like a sore thumb" and "too good to be true" LOL!

The only thing I am in two minds about is the possibility that I could have misinterpreted these events and that G was changing the way it assessed inbounds - a subject still niggling some of us today after the last change in displayed/acknowledged ibls.

I suspect the hard evidence that you seek, if it exists, would be chargeable and you must either do the research yourself or listen to the opinions of industry experts as I do and make your own mind up. If you do find any quality analysis on this I for one would be very interested. Good luck!

This "statement" is like virtually all statements in SEO/M it is an opinion, but one borne out by my experience which I thought I would share with Daria.

Thanks for sharing the info, youngpup. A lot of it boils down to experiment, then wait and see. Definitely right on about SEO/SEM "statements", there may be some factual analysis in some cases, but much is opinion and speculation (including myself) from experiences in SEO land.

Looking back it shows my lack of common sense after being given good advise from top players, SEOers who ride out these major shifts without making big changes.

Sometimes waiting it out to see what happens works the best. It may be safe, it may be boring, but it often works just fine as a strategy. Moderation (I know, yawn!) is really a good way to go when optimizing.

>Sometimes waiting it out to see what happens works the best.<

Is that really what some of you tell clients?????? :confused: I don't mean to be facetious, I really am curious.

What I tell my clients is if you want top 10 for DVD and you want it fast,then we better get a sh**load of links that says DVD. If you want top 10 for CHEAP DVD, that is not the same thing and again, you are going to need a sh**load of anchor text that says cheap dvd's.

Beyond that little concept, In my opinion, THERE IS NO PENALTY FOR ANCHOR TEXT.

It has been proven with every example of a google bomb. It was proven again with the win an I-pod for top spot for nigritude ultramarine, (btw, really great PR move Chris. Wish I had thought of it >nominates Chris Ridings for best PR move of 2004<). It is proven everytime you reverse engineer the bl's of the top 20 spots for credit cards, casino, phentermine, to mention a few.

Do a search for news. Ask yourself honestly if you really believe any one of those top 10 sites might not be there tomorrow because they got too many links that said NEWS in the anchor text. If you believe those sites belong there and you know they have a lot of links that say news and they are in the top 10 for news, then surely you can accept that your site would not get penalized for over-doing duplicated anchor text. True, those sites have a lot of links and many of them say something different, but you would also have to accept that they have a HELLUFA lot of links that just say news. Do you think any of those guys are worried about having one too many people linking to them with news as the anchor text and screwing up the percentage? Then why do you?
>no need to debate this specific example. As in all things SEO, it is about comparing apples to apples and this was used only to make a point<

I would think the industry would embrace the concept as a way to increase profit margins and revenues while being able to keep prices low. Stop telling your clients they have to reduce the relevancy of one link by altering the anchor text of another and start telling them to tell you what SPECIFIC term they want a top spot for. Dvd or Cheap DVD. If they take both, you can charge for both. It costs extra because you need about the same number of links for each term but they can save X% by ordering them both at the same time.

Million dollar hint. What is the adsense selling on the page you are getting the link from?

>In my opinion, THERE IS NO PENALTY FOR ANCHOR TEXT.

Agree, though I'll hedge a little bit by saying "no significant penalty." If there were, no resort or destination sites would be listed because 90%+ of the backlinks just give the name, i.e., Visit <anc>Podunk</anc>.

>Sometimes waiting it out to see what happens works the best.<

Is that really what some of you tell clients?????? :confused: I don't mean to be facetious, I really am curious.

I'm talking about situations like the Google Florida update. I didn't run out and change everything, but waited. And yes, at that point I did advise my clients to hold on because no one knew what was happening then.

I didn't know I was part of a "some of you" group. ;)

youngpup,

I think you may have taken my post a bit differently than I meant it. I certainly didn't want to downplay your opinion or experience. Rather what I wanted to do was use your post, to generate discussion of the wisdom of SE's possibly imposing a penalty for anchor text keyphrase saturation. Like you, I know many who offer the opinion that there is a certain threshold. Others don't agree. What I was looking for was to spur a general discussion in this thread as to whether it made sense for SE's to impose a penalty, how they would go about analysing links and explore some issues in plugging this into the algo. Here are some of the questions I posed:

1.Would the analysis only kick in once a certain number of links had been indexed? For instance, what if a site had 3 ibl all with the same anchor text, is that enough links to reach analysis stage?

2. Would a penalty or dampening factor be applied to links or to the site/page itself?

3.Why a 50% threshold?

4. Would the analysis differ for one, two, etc word anchor text phrases? For instance, a seven word anchor text phrase occuring 50% of the time may be more suspicious than a one word phrase.

5. If you had 1000 ibl's with the anchor text "www.mysite.com" or "click here" is there a justification for receiving a penalty or dampening effect?

Added: I'm also interested in the implications to branding that such a penalty might have.

Any thoughts guys?

The standing out like a sore thumb analogy has been made in reference to many areas of seo. As Massa indicates there sure are a lot of sore thumbs high in the rankings. :)

Jon

Jon, Thanks for your thoughts, I look forward to any further discussion. Although I know I am "lightweight" compared to these guys, maybe I was being fed disinformation as, after all, these are my competitors. I have wondered this, but put it down to paranoia.

Massa, thanks for your input, got me thinking again.

The question bugging me now is why sites with the keyphrase in the company name did so badly.

The question bugging me now is why sites with the keyphrase in the company name did so badly.

Let's assume there is a threshold and you've reached it in such cases. Doesn't that strike you as an unfair penalty? After all, you're only extending your "brand" name in a logical manner. Or others are responding to your brand with naturally occuring link text. Would a factor in the algo penalize that? If I'm Home Depot and everyone links to me with <a>Home Depot</a> where is the rationale for a penalty?


Jon

Let's assume there is a threshold and you've reached it in such cases. Doesn't that strike you as an unfair penalty? After all, you're only extending your "brand" name in a logical manner. Or others are responding to your brand with naturally occuring link text. Would a factor in the algo penalize that? If I'm Home Depot and everyone links to me with <a>Home Depot</a> where is the rationale for a penalty?

That makes sense to me, Jon. All sorts of companies would use their brand name and have that name in natural links pointing back to their websites.

Nice post youngpup, thanks for sharing.

>If I'm Home Depot and everyone links to me with <a>Home Depot</a> where is the rationale for a penalty?

The point is that the "real" everybody wouldn't. Ask yourself this question, how many sites link to dell.com with just the word computers?

Imho I think many people are going wrong purely in the way they are looking at things.

If you try and imagine things in the context of a pure algo you will only get confused.

As an alternative imagine an algo with filters bolted on here and there, some of the filters contridict themselves, some of the filters magnify other filters when that wasn't indended. Imagine a system held together by virtual duct tape where even the most knowledgable of those working within the SE have now idea what will happen when they re-index. Letting your mind wonder within that chaos is a better stratagy than trying to fit illogical square pegs into logical round holes.

As an alternative imagine an algo...

That makes more sense. Following the outcry after Florida I can imagine someone being given the job of damage limitation and their own jumbo size roll of tape with a brief to keep it strapped down until after the IPO.

link text is huge. :)

likely the #1 factor for many competitive phrases

I'm with ya on that Aaron. Anchor text upon authoritive sites is the #1 affect upon competitive rankings. Quantity means nothing, compared to quality in the competitive field.

I am just about to commence some testing on this aspect. It will measure the difference in rankings between anchor text from a relevant page, anchored to match the testing page. I will record a ranking, then place the same link anchor upon about 50 same site pages to measure if the engines do place any actual real ranking affect upon this technique and how much. Obviously ranking fluctuate, so any small difference would be a big zero, but a significant difference will prove the theory one way or another to how each of Google, Yahoo and MSN take this into account within their algorithms.

If you try and imagine things in the context of a pure algo you will only get confused.
Often seems the case for me. :confused:

Letting your mind wonder within that chaos is a better stratagy than trying to fit illogical square pegs into logical round holes.
When you say chaos I think non reproduceability and randomness. And that says no work for seo's. In that context I disagree that there is chaos to the rankings. However I don't think you mean it like that. I do like your description of bolt on filters and duct tape. But I disagree with your conclusion that because of that a systematic, logical analysis is fruitless. Maybe I'm just more naive in believing that Google cares, examines possible implications of and monitors results after algo changes. Certainly a reindex can cause surprises, but I think they try to be vigilant to those. I truly think that they want to minimize/eliminate unintended consequences from those bolt-ons.

The point is that the "real" everybody wouldn't. Ask yourself this question, how many sites link to dell.com with just the word computers?
But the real everybody may link to dell with<a>dell</a> far more than 50% of the time. Or for instance, Google itself. I suspect that they hit well over that level.

Jon

>However I don't think you mean it like that.

You know me well :)

The point I'm trying to make [not very well it seems] is that as a base algo matures it gets messy, chaos starts to creep in. To me chaos isn't randomness, there is still a pattern there just not one that can easily be seen by traditional thought patterns.

>Maybe I'm just more naive in believing that Google cares, examines possible implications of and monitors results after algo changes.

I agree, but it gets to a stage where it is out of control. Let me give a real world example, I can't find the info so this is from memory [I think it was originally posted by Tedster at wmw].

There was an interview with some search engineers from Northern Light [some time ago :)], they said they had no idea how any index would turn out, it was too complex. As an example they said they had a part of the "algo" that recognised that old established sites may be of greater value than brand new ones. They also had a part of the "algo" that recognised new sites maybe had more current info and maybe of greater value than old sites. ;)

>And that says no work for seo's

Thats where we will disagree. If it was as straightforward as "old" sites do better than "new" sites there would be no need for SEO's. To work with a system that boosts both old and new [as an example] is challenging, only full time professionals will be able to judge where the sweet spot lies.

>But the real everybody may link to dell with<a>dell</a> far more than 50% of the time.

Think about it.

>bolt on filters and duct tape

Anyone who has worked on an on-going project knows that even the best-managed ones become spaghetti code over time. The more complex the logic and, in particular, how many people join and exit the development team, the more tangled the code becomes. Like NFFC, I see no reason to think that G is immune.

Thanks for your sharing your philosophical approaches to examining se's algos. I agree that Google absolutely is not immune to complexity and unintended consequences. But again, I guess I feel they "examines possible implications of and monitors results after algo changes." And I do hold out hope that an analytical approach is not a useless endeavor. Geez, I don't even know if it's in my makeup to break down the algo any other way. :)

Jon

>your philosophical approaches to examining se's algos

Ouch that hurt!

Look at it another way.

To produce a great hit record all the power of technology is needed, all that crazy tech stuff you see when you see a recording studio. Anyone can buy that stuff, what it really takes is a hunch, a feel, a guess which is laid on top of the technology. That how I see SEO.

My turn to dig back.

>>But the real everybody may link to dell with<a>dell</a> far more than 50% of the time.

Have you thought about it yet?

No dig intended. In no way was I suggesting you don't also examine the algo in a structured manner.

To produce a great hit record all the power of technology is needed, all that crazy tech stuff you see when you see a recording studio. Anyone can buy that stuff, what it really takes is a hunch, a feel, a guess which is laid on top of the technology. That how I see SEO.

The huge difference is that there is a subjective element in your analogy - the human purchaser. Now perhaps someday even that will be quantifiable. There is no subjectivity in a search algo. Just math. You can subjectively interpret it but that has no effect on the actual function of the algo. However, the subjective interpretation of the music purchaser intimately effects the function of music sales. I don't discount a feel or hunch. I think that's healthy as well. As long as it's in concert with research and analysis. And that's why I think questions like those in post 11 should be asked. Even if some can't be answered. Or answers may not appear outwordly rational.

Have you thought about it yet?
I give up. My very unscientific analysis shows that 9 of the first 10 sites linking to Dell use Dell in anchor text. The other uses an image.

Jon

>9 of the first 10 sites linking to Dell use Dell in anchor text

How many use <a>dell</a>?

All, as "www.dell.com" and "dell computers" includes the keyphrase "dell". So examining keyphrase "dell" using my thorough research we find it occurs in 100% of text ibl's. Now what I was examining earlier is whether a subsequent devaluation/penalty of links, site/page or serp's (search for "dell") should occur or is occuring. And trying to find a rationale for supporting one. Your thinking seems to suggest that character to character link text matching may have to occur for unnatural appearance to arise. Then <a>the big red computer store chasing trains</a> is naturally different than <a>big red computer store chasing trains today<a>. Would you then discount a possible penalty/dampening for keyphrases that are a subset of link text? Or do you think one should only use that keyphrase 50% of the time as suggested in post 2.

Jon

>All, as "www.dell.com" and "dell computers" includes the keyphrase "dell".

I'm not disputing that.

The questions is:

How many use <a>dell</a>?

Perhaps I was a bit loose with my html coding for <a>dell</a>, <a>google</a> and <a>Home Depot</a>. What I meant to suggest was that keyphrase is part of the link text far greater than 50% of the time.

Jon