Today I want to touch on two issues that, when improperly perceived, can negatively impact how search marketers and site owners build out their campaigns.
Links to Pages & Domains
Many folks new to search marketing, and even some with experience, incorrectly assume that many elements of the PageRank model are still in place, including the idea that the link weight, anchor text relevance and trust passed from a link influence only the single page targeted by that link. In fact, the reverse is true - a link from a high quality, trusted source can raise the rankings of many pages on a domain because the search engines aren't just counting the link as a vote for the singular page, but as a vote for the entire domain.
This interpretation contributes to what I often call the "Wikipedia Factor" - pages at Wikipedia that have no external inlinks and only one or two links from Wikipedia pages will still rank in the top 5 for competitive queries. This holds true for domains like About.com, Technorati, Expedia, Del.icio.us, and many others - the cumulative link weight and trust that these domains have built influences all the pages they host. It's a big part of the reason why the Texas Hold 'em SERPs are filled with pages hosted on .edu domains, too.
A good example of this factor in effect might be last week's Anatomy of a Super Digg, though it's worthwhile to point out that Jennifer Laycock over at SearchEngineGuide feels strongly that Google and the other engines may remove the value of the inbound links when the topic is unrelated to the site's primary niche. I'm not sure this is accurate - in the offline world, companies constantly conduct promotions and engage in marketing that has little relevance to their products and services, simply to gain branding and mindshare. I think it's more of a philosophical debate than a sure thing as to whether the search engineers think that off-topic viral marketing or media attention unrelated to a website's core content is grounds for a lower weighting.
BTW - This "rising tide lifts all ships" effect is one of the biggest reasons to keep as much content as possible behind one domain, rather than splitting into multiple domains. I've also seen cases where subdomains appear not to receive or pass the full measure of this effect, which is why I always recommend that sites refrain from subdomain use unless there's an especially good argument for them.
Temporal Link Analysis
In many competitive ranking arenas, it is not enough to simply acquire the links necessary to rank well, you also need to continue to attract links to your domain over time, or risk being beaten out by a site that does. Sometimes, when it seems that the search engines aren't always giving the advantage to the site with the most or the best links, temporal link analysis is at play. Google, in particular, is very smart about looking for link growth patterns - that young upstart website that's moving up the ranks may surprise you because it doesn't appear to have the link strength to rank above others, but if you look at the pattern of inbound links, you might see that the newbie is attracting 4-5 new links a day, while the stodgy old timers are only getting that many each month.
A good way to see link growth over time is through a Technorati search (i.e. SEOmoz). Note how Technorati will show not only the number of links coming in, but the timeframe - 2 hours ago, 3 days ago, 4 weeks ago, etc. Since a lot of the fastest link growth happens in the blogosphere and other sites that issue RSS feeds (and thus can be tracked by Technorati), this becomes an invaluable tool. I've always been immensely frustrated that none of the existing web analytics programs offer a feature that allows you to track new inbound links over time (based on URLs that have never sent traffic in the past).
BTW - Next week should be very exciting, as our premium keyword research article launches and a new free article on sources for effective social media marketing (courtesy of Jane) kicks off, too. Plus, Rebecca will be giving her very first public presentation at the Portland Searchfest.
Rand, and what about the reverse impact? What if a link is removed/test ad discontinued or any similar thing? Does that lowers the PR and in turn the rankings?
A site that got upto delicious/popular or digg home, may have produces a masterpeice resource but what about the duplicates? Can't G discount the link value for a period of say 20 days after the sites that went upto digg home. And then count for if a resource from the same site appears in digg later in the month/year/anytime in future? Is that possible?
Excellent article 10/10, trustrank is key, I have a very difficult time trying to explain this to some of my clients who know just about enough for it to be dangerous. I am starting a blog on my site and I was going to use a sub domain but I wasn't sure if it would have the trust of my site or not but reading your article I think I will try and set it up domain.com/blog as apposed to blog.domain.com.
Rand, this is your best post in a while.
I agree with everything you said except for the subdomain issue.
>>Many folks new to search marketing, and even some with experience, incorrectly assume that many elements of the PageRank model are still in place, including the idea that the link weight, anchor text relevance and trust passed from a link influence only the single page targeted by that link.
Well, there's a straw man if ever I saw one. The PageRank model, as you refer to it, never did refer to the value staying on one page but specifically included iterations where the PageRank would flow around connected pages.
I very much doubt that you ever believed that PageRank was "static" - and if you did I am extremely surprised - and would be very interested to hear of the "experienced SEOs" who did (unless they are these "non-expert" ones who appear to be getting so much press these days!).
There is certainly a valid point in what you are saying in that domain integrity has moved up in importance in both ranking and linking (which means that the relative value of links often has less to do with PR and more to do with their provenance or destination).
Regarding temporal analysis, surely if a "young & thrusting" domain beats out an old authority there is also the implication that that young domain will have to keep acquiring links at the hectic rate that is keeping it above the old authority or risk sliding below it again (or at least until its own authority kicks in).
Rand I've been watching "hotspots" on 14thC where one post gets some buzz & backlinks pulling the neighbering posts up with it. This is part of why I think PR may be more valuable than I once thought but PR explains this better than a blind domain grace. The PR transfers from page to page through links getting a bit weaker at each junction.
And then through the main navigation some of the overall "link strength" just naturaly filters throughout the site.
Another example of this is linkers union where one really good link hits the domain but two neighboring posts get several strong links. The site is still small enough to see pretty clearly how the juice transfers from one page to the next... and divisions of PR are very evident.
Your theory makes me think that any site that gets enough authority links won't have to worry about suplemental results. And I don't think that is true. At least I have not witnessed it myself even on powerful sites like Wikipedia and About.
My $.02
The way I've seen it work is as a two stage thing. The domain, like you've said, passes weight down onto each individual page, with the level of weight getting lower as you pass each degree of seperation.
However, as far as I can tell, you need to actually have a certain link weight to that page for it to come out of supplimental, which acts like a trigger. Quick example...
A domain has 100 points of links. A page three degrees down has 20 points, from the domain. However, it has no links pointing to it, which triggers the supplimental filter, and therefore it doesn't show. It's not deemed valuable enough.
Another page three degrees down also has 20 points from the domain. However, it has 30 links pointing at it, and so the supplimental filter isn't triggered. Thus is shows up.
That's how I see it anyway.
Your take makes sense (right with you on the seperation anxiety) but only if the value of a page is relative to other pages within the site and not necesarily all pages in Google's index.
Ah, screw it - just get some deep links and forget about it!
So you're saying that it works like gravity, with every page exerting some weight on every other page, simply by its existence?
The way I see it, every site is a web unto itself. The links that it has pointing to it are then link strands of another web, attaching it to something bigger.
That way you have several levels - the page, which is the smallest part, which is part of website, the next sized web. Then you have the local cluster, which is the websites that link to that website.
Personally, I'd have stopped there in the past. But whilst I've been writing this, I've realised something. Moving out, you have the outer cluster, which is all the sites that link to those.
Everything expands out from that. So you end up with butterfly effect, where a webpage somewhere else will have an effect on every other page (unless it's got Nofollow).
Never looked at it quite like that before. Cheers!
I still think a page needs a certain number of links pointing at it to rank though, irrespective of the domain's weight. Either that, or there's some other trigger. Which begs a question:
Could you bring a page out of supplimental using something other than links? What do you think?
Not gravity by virtue of existance... think of it more like a virus. Each handshake (link) makes the virus a little weaker but some of the effects still get through.
You can get pages out of the sups without links if the problem is duplicate content or not enough content just by adjusting the text. If the issue is not enough oomph then you do need more links but not necesarily to the page itself but to one that links to it. For example I noticed the press releases on 14thC were in the sups. So I got a couple links pointing to the topic page and that got all the sub-pages back in the index.
"It's a big part of the reason why the Texas Hold 'em SERPs are filled with pages hosted on .edu domains"
huh? on Google?
Great post.
I have a different view on the subdomain issue. I usually make that decision based on the number of pages the site has, how many subdomains I might want to use and how they will all interact. It is definitely different for different sites - this isn't a cookie cutter decision.
And I absolutely HATE when people refer to page rank like it was 2004....those misconceptions are the hardest to try and explain to clients who "read in a magazine..."
I've always believed that an inbound link only benefits the particular page that it is pointing to. Although you have me thinking now (hard sometimes) and I guess there is of course a trickle down effect happening so in that sense the link could benefit "the entire domain." Anyhow, very insightful post...as usual.
About.com has all of those subdomains that seem to dominate the SERPs.
keyword.about.com
Do you think they are just getting their link juice from the main domain linking to them? or is it all of the cross linking they do between the subdomains?
I think it is a combination of the two. I have a client who still uses subdomains for many portions of the site and links across subdomains and also from the homepage.
This client has great rankings for many terms and while I don't think it will work for everyone, I have seen search results where 9 of the 10 first page results in Google come from the different subdomains for this site.
Of course this site has been around for a long time, so I don't think the same would be true for someone trying to launch a new site in the same fashion.
thank you for information
You mention that "juice" does not flow through sub-domains as readily. What are your thoughts on that same "juice" flowing through directory levels?
Sub-domains are obviously a separate domain so even if you put a directory at sub-domain.domain.com, it's going to have the same adverse effects as any other sub-domain would.
given: www.domain.com or www3.domain.com, two completely separate domain names. The world concludes that www is the default so that is what everything thinks to use. People don't even realize that it's a completely separate domain name.
I use the enforce-www method to force non-www users to my www.domain.com instead so even all the link "juice" gets put on the www.domain.com instead of both the domain.com and the www.domain.com
thank you for information
I doubt if A backlinks to B then SE should consider that A trust any content in B
is A trsusted enough?
is the link down?
is B old or new?
is the link descriptive and user friendly?
does the anchor text describe the main keyword phrase of the content on B?..
AS SE thinks the vote is a vote for the page not for the whole domain.
AS persons, maybe, it depends on the culture and how people think, so indirectly this could be useful...
I thimk that what I think right now is fair enough, i don't know if there is an excepion.
Very enlightning. I wasn't aware of the wiki factor to this extent. So PR spills over to the entire site, but what about the relevance of the inner pages, does that play a role on deciding which pages get more PR from the link?
No. PR has nothing to do with relevance, just link power.
...all of which accounts for the overbearing power of domain over pretty much every other factor. Once a domain has that strength you can rank for all sorts of things not related to the core of the site - which can be a blessing (lots of traffic you can pass on to a client perhaps) or a curse (lots of one page visitors).
Interestingly, I'm seeing some non-domain-relevan-but-ranking-highly-thanks-to-domain-strength pages starting to get weeded out of the SERPS. I think Google *is* playing with overall domain relevancy when assessing single page rankings. Be interesting to see how that unfolds...
"In fact, the reverse is true - a link from a high quality, trusted source can raise the rankings of many pages on a domain because the search engines aren't just counting the link as a vote for the singular page, but as a vote for the entire domain."
...partly because PageRank flows from insertion point to other pages on a domain via internal links.
"This interpretation contributes to what I often call the "Wikipedia Factor" - pages at Wikipedia that have no external inlinks and only one or two links from Wikipedia pages will still rank in the top 5 for competitive queries."
Wikipedia ranks well because all its internal anchor text contain the exact keyword and millions of links pointing at the site making its pages both relevant and important, the two basic ingredients Google uses to rank a page. Not to mention people also often use the target keyword to link to Wikipedia (e.g. "tomato" to link to the "tomato" page).
Title tag, onpage text and internal links all say that the page is about "tomato." Anchor text used in links from other sites also says the page is about "tomato." Thousands of IBLs say the tomato page is important.
https://en.wikipedia.org/wiki/Tomato
TBPR 6 isn't anywhere enough to secure a #1 ranking but combined with a high relevance score, page value seals the deal.
I think this makes a pretty logical analysis and interpretation by SEs, otherwise, they would have to question the likelihood that an authority site would link to a great page on a completely dodgy site.
What about the value of PR on a low PR page from a high PR site? Say XYZ.com has a home page with PR8, but the page with the link to your site is a PR1 (directories are a prime example).
Any thoughts on the weighted value that is passed? I don't think we can expect Google to say, we'll give you the same value as if the IBL was from a PR8 page, but it probably carries more weight than a PR1... but where within that range does the value come in at?
Of course, there may be too many other variables to ever truly say.
Well, this is the point that I was trying to make earlier. If I'm looking for interesting sites to link to, or targeting content which would be attractive for them to link to, then these days, I tend to ignore visible PR and backlink numbers and look at the quality of the site and the neighbourhood.
I would take a link on a low PR page from a domain with integrity (and by that I mean "integrity in the eyes of the engines" as well as integrity of content and message) any day over a link from a higher PR page on a directory that was full of SEOs and their clients.
As soon as I see web design & hosting links, Amsterdam hotels and printer refills on a site, I'm out of there.
You mean those are relevant?
So I suppose you would immediately leave a mortgage site that had AdSense ads and links for ringtones and free Flash games in their footer?
Site snob! ;)
I like your graphs. I'm a visual person. even though i'm a writer, i'm not great at reading. so graphs are good. very very good.
Good points Rand.
I've always been intruiged if inbound links carry more/less/any weight depending on the actual type of document i.e. HTML vs dynamic ASP/PHP/etc. I've always held the belief (possibly wrongly) that static HTML pages, through the nature of not having dynamic variables in them, tend to pass links more easily than dynamic ones.
Does anyone have any hard eveidence if this is true or not?
No hard evidence but how about a lack of evidence? I've seen no difference between .html and .php. I prefer to hack my htaccess and show .html but that is for asthetic and security reasons - not SEO.
Agreed. I've never seen anything to show this.
I doubt this can be true. Don't forget anybody can make a dynamic php page look like a static html page quite easily.
I wouldn't think the basic nature of static versus dynamic should matter.
But I think what you are really after is the quality of the URL.
In other words, if the URL is a really bad SEO URL with tons of variables, session IDs, etc., would the SEs consider the link less because they can't be sure of the contents of the page.
I think that is an interesting consideration, more in line with what you were wondering, and a very different question than just static vs. dynamic.
I agree. Just look at this website, web pages don't really need any "extensions" and SEs would be smart not to take into consideration of something non-standard. If we are comparing .php vs .html, what about .htm vs. .html? :)
I guess that's something I should have added to my initial comment and it's a very good point. . I look at some URL's with tons of variables that make the URL about 100 characters long and think "How is a search engine supposed to fathom out what is going on in there". I wonder if search engines downgrade the importance of links if there are too many variables in the URL?
Of course, if you're using 'post' data then it's harder to get access to that.
The only evidence of links quality to HTML vs. PHP pages that I could see is the quality of a link's URL. If a link to dynamic page has too many variables, that could lead to problems. Other then that, it shouldn't matter.
Very interesting including the sub domain argument.
If inter linking passed on PR ove can not the subdomain get a bit of that loving?
In my experience I have a new client with more than 250k incomming links from poor sites and i trying to get visits from google but I can´t cause all those incomming links are crap.
Google only shows 7 incoming links while Page-strengt shows the 256000.
Note: please don´t try to get incomming links with the same word, that is a hard mistake to fix.
Try using Yahoo! Site Explorer to check for backlinks
Ahh...Looks like the "google bomb" fix might be hurting you a bit....and you are right, it's VERY hard to fix.