February 2012 Google Update Affects Link Evaluation

A Link Evaluation Signal is Gone... Which One?

[martinibuster]

According to this Google Blog Post:

http://insidesearch.blogspot.com/2012/02/search-quality-highlights-40-changes.html

Google has removed one of the link-related signals of what the topic of a page is. Understandably they didn't say which signal it was:

Link evaluation. We often use characteristics of links to help us figure out the topic of a linked page. We have changed the way in which we evaluate links; in particular, we are turning off a method of link analysis that we used for several years. We often rearchitect or turn off parts of our scoring in order to keep our system maintainable, clean and understandable.

So what are the signals related to the topic of the linked page? Here are some suggestions. Feel free to add more.

Title Tag
The title tag is a signal of what a web page is generally about. The title tag has been used to help identify the meaning of a linked page from the general theme of the page linking to it. The title tag is a signal of the general theme of a web page. The topic of a link can vary from a side topic to a more granular or a completely off topic meaning.

Surrounding Text
This establishes the context of a link, thus helping to define what the linked web page is about. This also helps identify if a link is paid for or is associated with a donation.

Position of a link
Where a link is located is an important signal. A link in the footer is presumed to be less important than a link within the body of a web page. Navigational links are presumed to be depreciated according to a set amount, perhaps more than outbound links but just a fraction of a normal non-depreciated link.

HTML signals related to weighting of a link
These might include Heading Tags [w3.org], bold, italics, capitalization and font size [webmasterworld.com] as factors for assigning greater weight to a link. Font size is an interesting candidate for depreciation. Here is what the orignal Google Paper says about Big Fonts:

Aside from PageRank and the use of anchor text, Google has several other features... Google keeps track of some visual presentation details such as font size of words. Words in a larger or bolder font are weighted higher than other words.

[edited by: martinibuster at 9:46 pm (utc) on Feb 28, 2012]

[Andem]

>> we had seen a decline in influence from cheaper links since january

Sorry if I seem a little bit forward, but aren't 'cheap', AKA paid links exactly what Google is trying to discount? That example is both flawed and irrelevant since paid links are blackhat and that isn't what this forum is about.

On the complete opposite and whitehat side of the universe, I understood that branded links to inner pages were exactly what Google prefers (and even moreso since Panda). In my case though, neither branded links nor more specific, more keyword-specific links which have made a difference. I'd guess that at least 3/4 of all links to any page on my site reference either just the brand name or the brand name combined with the page topic, all organic.

Since I've heard about the newest updates to their (Google's) rankings, I haven't really seen any changes with regards to traffic, though as with almost every "Panda" iteration, traffic is just that little noticeable bit less targetted.

[martinibuster]

First, I would discourage anyone from defining what this forum is about. The definition is here, in the charter [webmasterworld.com].

The example is not irrelevant. Regardless if they were cheap or paid for, it's a signal that the member reports is not working. This is important.

[edited by: martinibuster at 12:38 am (utc) on Mar 6, 2012]

[Sgt_Kickaxe]

There is a lot of over-thinking going on here(and I do appreciate that, all ideas remain on the table) so I tried to take a step back and look at the bigger picture. It's apparent that Google has gotten better at evaluating what a page is about regardless of incoming links and I would be willing to wager that they now trust that evaluation process a lot more. If that's the case it would make sense to turn off incoming link text, e.g. anchor text, as a ranking factor.

It makes sense that once you can reliably understand what a page is about that you do not need other pages to tell you what they are about. From a search perspective the incoming link text CAN BE MANIPULATED and as such is/was probably high on the list of signals that cannot be trusted.

Just my 2 cents, I think relevancy of links and authority of site linking to you still play a major role but actual text has probably been replaced by something else, perhaps tone of content? If a page that links to you is positive or critical in nature is probably a better ranking factor, especially if Google places value on visitor experience(who wants to land on negative sites?), than anchor text but that's for another thread.

[seoskunk]

I would say title tag has been depreciated, and it makes for much more interesting results .....

[tedster]

Might be nit picking here but this was not part of the Panda update, but part of 40 other changes in the early March update.

I completely agree. Discussing algorithm changes at Google is becoming chaotic because webmasters seem to be calling any part of the algorithm "Panda". It's not, and this change was clearly distinct from Panda.

This is not nitpicking at all, IMO - it's about clarity.

In the past, Google has used many signals from the linking page to get a handle on the topic ofn the target page. Not just anchor text, but also "nearby text" and even the linking page's title element. IMO, this is one area where the dials were turned down. They found more accurate ways to judge the topic of a page than looking at these relatively remote signals.

At the same time, it seems that a lot more sites lost rankings (in a penalty-style fashion) for too much matching anchor text. It's like a new wave of "over optimization penalties."

[Sgt_Kickaxe]

At the same time, it seems that a lot more sites lost rankings (in a penalty-style fashion) for too much matching anchor text. It's like a new wave of "over optimization penalties."

Some sites use an automated "similar posts" or "other links of interest" section that leads to other pages. If you're right I wonder if internal links always repeating the same text is a problem. I can't see Google ignoring that fact.

[CainIV]

If that's the case it would make sense to turn off incoming link text, e.g. anchor text, as a ranking factor.

That's my hunch too.

If it didn't get set into motion this time around, it will soon.

[idolw]

There is no anchor text in social media links like twitter and facebook, and google counts them and attributes proper topic to them.
In fact, getting rid of anchor text as key ranking factor would make it way easier for non-professional SEOs.

[Sgt_Kickaxe]

exactly idolw, the Google article mentioned wanting things less complicated and ignoring anchor text would give them that, as you mentioned many links don't have any and not all webmasters know enough SEO to use them effectively.

My vote is on link text, I'm trying to figure out a way to effectively test that - suggestions?

[danwhitehouse]

Just spent a few minutes reading this entire thread, it's fascinating to see different perspectives. So here's the things that I've picked up on from the press release:

1. Link evaluation
2. 'Topic' of a linked page
- Topic is a very specific word. It's not content, niche, industry - it's topic.
3. Turning off
- Turning, future tense.
4. Several years
- Could be a general number of could actually refer to something they have used for 12 years
5. Rearchitect
- They feel that they can not re architect this signal, so it's going off.
6. Maintainable, clean and understandable
- What would create so much information in a system (that Google feels makes is) to create so much clutter that they feel they have to 'clean'?

Google turned off meta keywords a little while ago. Bing did the same, but decided to turn it back on again. Maybe the clutter is leading to meta information as a whole. I can see how it has been used for 'several years' and does in fact create quite a lot of clutter if you think about it. Search engines are creating their own titles for pages more and more frequently making the use of title tags redundant.

With search engines placing less benefit on 'seo', so to speak, and more on content and social media, I can imagine that this is exactly the type of thing that they want to move away from, and they have been for some time.

I also think PageRank could be a viable candidate, but why do they keep on updating it so frequently? Maybe to get one last data set before pulling the plug? Thinking about it, wasn't the original patent application filed by Google all about Page Rank? Surely 'Page Rank' is just a public relations term to keep all the oldies engaged with Google due to the history and association between the 'term' and the company. I call it a term because (as we all know) it has now bearing on the modern web.

What is something that Google does not attribute trust to

I've been doing some searches from 10+ years ago to try and get to the bottom of this. Here's a w.tools thread that talks about EDU links being important.

[webmasterworld.com...]

Those are my thoughts, back to work for now.

[danwhitehouse]

Further, I just thought of an analogy for the removal of anchor text (if they do so).

Does anyone remember back in the day when, if you wanted to install a new piece of software on a pc, or install a sound card for example, you had to go into the code, free memory, change setting etc etc. The list was as long as my arm.

Now, everything is colour coded on the back of the machine for peripherals, and plug and play exists with auto installation and so on.

I wonder if 'SEO' is going this way - so easy that anyone can do it.

[jkwilson78]

To add some more fuel to the anochor text debate:

I have a site that I pretty much abandoned a few years ago. I noticed 2-3 months ago that it had started ranking on page 2-3 for some of the main keywords in the niche.

So I decided to have some fun and get some cheap anchor text specific links....lots of them....junk links....on off topic pages....

As a result rankings have shot up to page one across the board for the keywords targeted by the anchor text of the links.

The most recent ranking boost has come since Google has made this announcement with some rankings going from bottom of page one to the top 5.

Either Google has not yet rolled out some of these changes or I really doubt anchor text is what has been devalued.

[BeeDeeDubbleU]

As I said earlier it may be internal anchor text.

[Planet13]

For what it's worth...

For my sites, it seems like neither external nor internal anchor text has been devalued.

[maddawg]

The websites I'm seeing drop from the rankings used low quality blog posts with keyword rich anchor text.

While there are still sites with a lot of keyword rich anchor text at the top of search results, I still believe there were a lot of websites ranked on page 1 who only got their rankings as a result of their anchor text that are now getting hit.

[scooterdude]

As a matter of interest, what do folk think Google gets out off this announcement? I'm puzzled


With the Above threshold announceement, clearly they wanted everyone to change the advert to content ration above the threshold, so announcement gets everyone modding frantically,

[idolw]

jkwilson78 - remember your old site is not in the vacuum. Your neglected site may have had a bunch of high quality links (high quality link does not mean it has perfect anchor text) that built trust to the site, and competition might have been basing on crap links only. So you did not go up but the competitors could have gone down :)

From my experience the best links rarely have the anchor text I want. Usually they are non-anchor links or just domain name links or feature not very targetted anchor text. But I am no expert at all ;-)

Again, at the beginning Google made their SERPs too powerful and have been fighting to make them less powerful for years. By powerful I mean that being #1 for your keyword was the holy grail. The fight with SERPs power can be seen by more and more ads, universal search, etc. so companies spend less on SEO and more on advertising.

All these changes try to make SEO even more random business. For a small webmaster like me, anchor text removal is a good idea. I can spend more money on good content and on-page optimization then spamming people for links.

[SnowMan68]

@ maddawg

I am seeing the same thing. Big drops on sites in the serps (short and long tail) using only the keyword rich anchor text links from blogs and spammy directories.

[edited by: martinibuster at 9:01 pm (utc) on Mar 7, 2012]
[edit reason] Fixed typo. [/edit]

[vrottenstein]

I'm also seeing big drops on sites with a link profile full of blog comments with anchor targeted keywords...

[Planet13]

@ SnowMan68 and vrottenstein

Big drops on sites in the serps (short and long tail) using only the keyword rich anchor text links from blogs and spammy directories.

I'm also seeing big drops on sites with a link profile full of blog comments with anchor targeted keywords...

Can you elaborate on what you mean by "big drops"? How many positions (or pages) were the ranking drops?

[vrottenstein]

Yep: 50 keywords on TOP 5 plummeted to 2nd Page or between position 7 and 20.... , and I have this scenario again and again in 3 or 4 websites ...

[mike2010]

why do they have to be weird & play this guessing game crap? Just tell us what it is & be done with it.

Now i'll have to check back on this thread in a couple months to see what the 'general consensus' is on what exactly was changed.

soooooooooooo stuuuupid.

[mhansen]

As a matter of interest, what do folk think Google gets out off this announcement? I'm puzzled

I think you nailed it on the head! Keep everyone puzzled.

[Planet13]

why do they have to be weird & play this guessing game crap? Just tell us what it is & be done with it.

well, then ALL your competitors would do the same thing, since it is so easy. Then you would be back to square one.

Don't wish for things to be easier. Wish for yourself to be smarter,

[Robert Charlton]

Great discussion. Just some random musings on this. I'm thinking they haven't substantially discounted anchor text....

...so anchor text says merely "this page" or (shudder) "click here"

I did a sloppy test of the anchor text theory by searching for [click here], and I see that the Adobe Reader Download page is still in the #1 spot, with another Adobe pdf reader page as #2. I'd attribute this ranking to anchor text influence, as it's hard to find another reason why the page would rank on the term. I can invent a more complex description, but it ultimately gets reduced to "anchor text".

The Adobe onpage text, btw, is "clicking"... a stem. Several other major download pages rank high up on page one as well, arguably for anchor text influence as well. Again, this is a very sloppy test.

I also feel that removing anchor text influence completely at this stage in the algo would be too disruptive. Google is more into evolutionary kinds of changes. Overly repetitive, untrusted anchor text is something else, but I assume that's been gone for a while.

I could see that Google might have reduced the effect for certain anchor text source-and-target-page patterns. Eg, some years back, we had Google bombing of GW Bush's bio for [failure] and [miserable failure] removed... but the White House bio reappeared for a short time for the query, as I remember, when the word "failure" crept into the bio text.

Now we're seeing [santorum] results changing, but fading in and out. I can't currently reproduce the results that Danny Sullivan reports [searchengineland.com...] ...and I suspect that there's ongoing battle with some filter thresholds. Conceivably, whatever Google had done to affect these results, at least temporarily, could have been the factor described in Google's top 40.

I was intrigued by the 301 possibility, particularly if the 301s were combined with something resembling Google bombing. Definitely something that Google would want to get rid of, if it could separate those from legit 301s of redirected sites.

I'm just coming off a site redesign project in which all pages were 301ed, and I've got to say that legit 301s, at least, seem to be alive and well, with the new site appearing for old rankings in about 4 days.

I also don't remember ever hearing about or noticing that 301s were depreciated. Matt Cutts had said that there was some loss of PageRank in 301s [webmasterworld.com...] ...but I never saw this loss as a depreciation. It seemed more like a loss inherent in the redirection process... or the math of the process... somewhat analogous to friction.

I should note that some, not all, 301ed domains, have been showing up in the serps if you search for them... since about June 2011. It's not actually necessary for Google to keep track of these old domains for 301s to work. Theoretically, the redirection is all done by the server and browser/user-agent. So, if Google were dropping credit for (some) 301s to streamline the system, one might wonder about Google's displaying these artifacts in the index. Here's the discussion....

Domain name replaced in SERPS with alias domain name
http://www.webmasterworld.com/google/4327200.htm [webmasterworld.com]

Thinking out loud here, the "title tag" might be a possibility, since Google is almost rewriting these at will, based on internal content and inbound links... but deep in my gut, when I look at rankings for certain pages, I know that they're strongly influenced by the title element phrasing.

Possibly, the brand name and its associations out on the web are carrying some of the information that used to be carried by keywords in inbound anchor text, but this, I'm guessing, would be a turning on of new factors rather than a turning off of old ones. The behavior of the title itself in these cases doesn't seem to be changing.

[Planet13]

Great points, Robert:

to chime in, the ONLY evidence I have (and it is entirely anecdotal) tells me that:

1) Anchor text is alive and kicking

2) Surrounding text has lost influence.

Again, this is purely anecdotal.

Someone linked to a page of mine and used the keyword I was trying to rank for in the string of anchor text (along with three other words not really related to the keyword.)

That page started getting a LOT of referral traffic from the linking page right away, and then started getting a lot of traffic from google serps after a few days.

The surrounding text immediately outside of the link is irrelevant to the keyword or the topic of my page.

Of course, this is just ONE incident.

[Shaddows]

That page started getting a LOT of referral traffic from the linking page right away... the link is irrelevant to the keyword or the topic of my page

Something else must be going on there. Regardless of the surrounding text, that link is either valuable to the sites users, or it isn't.

You say it provides traffic, so it must be useful. If it's useful, Google should be rewarding it.

Also, it is illogical to infer that surrounding text is irrelevant. At best, you can infer that poor context does not 100% negate link power, which can be restated as "context is not the only ranking factor affecting links" - and that is hardly revolutionary.

[Marketing Guy]

It's not actually necessary for Google to keep track of these old domains for 301s to work. Theoretically, the redirection is all done by the server and browser/user-agent.

True, but wasn't it the case for a long time that spammers would buy up expired domains and throw some content on them based on previous subject area to leverage the old domain's "SEO history" (for want of a better term) within that niche?

That's a separate metric from PageRank (essentially qualitative vs PR's quantitative nature). It would be possible for Google to maintain the history of a domain (in terms of subject area) quite easily and this would be useful to flag sites which change dramatically.

Looking at the precise statment from Google;

We often use characteristics of links to help us figure out the topic of a linked page.

If there is a qualitative metric associated with a 301 redirect (i.e. Google uses it to pass the subject area from old page to new page), Google could switch that off with minimal ranking distruptions for legitimate use (i.e. PR is still passed, or at least a percentage of it). New page would still rank as well as it otherwise would (it's optimised and is receiving most of the old PR), but completely changed pages wouldn't necessarily be able to achieve the rankings of the old URLs (they are judged on how they are now rather than how they were).

May muck around with a test on this one when I get some time.

[SnowMan68]

@ Planet13

I guess it was more of a big drop in overall traffic for a lot of the sites that I was monitoring in my niche. While they dropped to pages 2 and 3 for most of their prominent short tail stuff, overall traffic plummeted 70-90 percent. I know that what I use to track the google referrals isn't exact, it does give a pretty good indication (positive or negative) how the site is doing in the google serps. From what I could tell the only link building being done was through blog posts, social bookmarking and directory links.

[CainIV]

So I decided to have some fun and get some cheap anchor text specific links....lots of them....junk links....on off topic pages....

This isn't conclusive. Maybe those links, by nature, are counting positively, and the anchor text is still being devalued. The only way to truly test - and receive some respectable results to work with - is to create a controlled experiment between two VERY similar domains.