It's true you can do a lot of work on a site to optimize your rankings, but eventually you're going to need some links. While they make small changes all the time, the search algorithms are still based primarily on links. But here's what you may not consider when you're devising your link building strategy or working to acquire the links you need: links can accomplish very different things.
Depending on your situation--your site, your link profile, your competition--the kinds of links you need and the things you need them to do for you can vary significantly. Learning how to determine the kinds of links you should pursue, and what they need to accomplish can save you a ton of time, money, and frustration. Watch this week's Whiteboard Friday to learn how to segment your link strategy and focus on areas of weakness.
I think this could form the basis for a new PRo tool.
Put in a page/URL and it returns you a matrix of scores for the 4 different factors.
I think so!
That would be so increadibly helpful!
:)
Just an add-on observation - There's one aspect of "Raw Link Juice" that I think many people ignore when they're just starting out, and that's the number of links on the linking page. Sources like low-quality directories and over-spammed dofollow blogs too often mean that you're just 1 of 500 links on that page. Regardless of overall link juice or even authority, how powerful is a link going to be when you're getting 0.2% of that page's juice?
I have always given importance to the anchor text on the inbound links from the SEO perspective.
Thanks , now I am more confident about it.
And yes, I can show this video to the clients when they tell me that the other SEO companies are giving them 1000s of directory links.
As I have identified just a few specific country specific directories and industry portals in which the sites are submitted for SEO linkbuilding purposes manually.
Moreover, as usual 'Content Is King' so quality content is like natural link bait. If the site is having good informative content then it automatically gets a lot of links from various sources.
Moreover, now with social networking people are sharing links like never before, which again is a good source for inbound links.
Next time I'll wait with my next sip of coffee until after Rand has explained his "little treat". Almost ruined my keyboard because of that Disney story :-)
Also to keep in mind the focus or campaign level of any link building. Especially as Rand mentioned comparing against others based on ranking.
How well these buckets are filled out may vary depending on this piont of view.
Are we talking global level... Big Brand or Industry?
What area of the site is being focused on... homepage, top level, interior?
Or are we talking about a more refined focus... a particular problem or solution, a subcatgegory, a particular product.
A site could be weak or strong in any of these areas, so it is important to look at it as more 3-dimensional and find the weak spots that way as well.
I'm defintely a believer of the power of anchor texts, and even in relation to internal linking I've seen the wonders it does, when used right. The second thing I always take into consideration is link juice. After all, everyone wants to get a link from somewhere nice.. I've never really thought too much about domain trust and link diversity but it's something I could always look into and have a play with, thanks for that..
I'm also all up for the tool which checks these various metrics, it'll make our jobs a lot easier and make us all more lazy, lol...
Some quick corrections - Adobe's PDF reader ranks #1 for "click here" - not their homepage. And Posterous.com ranks on page 1 for "via email" (due to their many links that have this in the anchor text).
@wakeless - I agree! It's in the works, but likely won't be out until late summer. In the meantime, you can hack it together just grabbing metrics out of OSE, using the API or even hacking together with the KW Difficulty output.
Rand, TGIWBF! (Thank God it's White Board Friday!). I enjoyed this one, but there's a missing piece for me here. What happened to relevance? Are you not a believer that links are more powerful if they come from thematically relevant pages / sites? I think I recall a post of yours a while ago that was skeptical about the need for this kind of relevance, but it seems hard to accept that it's not important. What is your current thinking about this?
I would say that relevancy is being assumed a bit. Hopefully most people, and most readers here, will incorporate relevancy in any link building activities they do as much as they can.
Relevancy can come in many flavors here though as well... relevant based on the linking domain, linking page, anchor text, etc. And are these relevant in a global sense, or ultra targteted, i.e., company or industry vs. a very specific topic.
Certainly there might be extremes of irrelevant linking, which might send flags for other reasons, but probably less of a concern. Rather, we might expect that the engines may simply give a little more weight to links that appear more relevant than actually ding any site due to irrelevant links.
Otherwise, others might engage in irrelevant link building against a competitor to ding them.
Identity, you raise some points here that have been much on my mind. How is link relevance determined? An idea that I read in this interesting YOUmoz blogpost post that was new to me is that
Internal Link Relevance = Linking Page Relevance x PR
External Link Relevance = Linking Page Relevance x PR x Weight (>1)
"Weight" is used here, obviously, because external links are given more weight thaninternal ones. Now, I would not neccessarily accept this opinion literally (it assumes to much about the algo), but I think it makes sense that the value of a backlink would be some kind of product of the linking page's relevance and PR (we have all seen how pages with a low PR can outrank pages with a high PR because they are more relevant).
But how does Google determine the relevance of a linking page? Well, I'd be surprised if the process were not highly similar to how Google determines the relevance of any page, i.e. by looking at the title tag, keywords on the page and other on-site factors.
Interestingly, the missing piece in the above equation is precisely the anchor text, which I think is a major omission. Anchor text cannot be subsumed under the linking page's relevance. Links from irrelevant or partially relevant sites contribute to rankings for a keyword thanks to relevant AT.
I am surprised that a good number of SEOs that I know think link relevance doesn't matter and that there is "no evidence" that it matters. I suspect this has something to do with a few things people have read at SEOmoz. For example, in this post Rand says: "I personally don't worry too much about topical relevance - if you can get a link from a topic agnostic site (like NYTimes.com) or a very specific blog on a completely unrelated subject (maybe because they happen to like something you published), I'm bullish that these 'non-topic-specific' endorsements are likely to still pass positive value." I don't doubt that the still pass value, but I don't think it's a reason not to worry about relevance as one of the factors in the calculation. Provided, of course, that relevance matters.
And here Rand remarks about the idea that "themed links" have greater value: "While personally, I've seen little evidence that an algorithm like this exists at Google, Yahoo! or MSN/Live (haven't honestly done enough Bing investigation to feel confident making statements around their practices), I'm very curious to hear your thoughts." BTW, I wonder what the results of that survey were.
Rand, what is your current thinking on this issue? (Would love to hear what others think as well, especially if they have empirical bases for judgment.) I'd say relevance is a strong candidate for a factor in addition to yout four. I am open-minded and would love to hear your position. I daresay an up-to-date look at this would make for a fantastic SEOmoz pos, and I'd be your first reader!
It's a good point, and potentially a fifth bucket, but...
A) There's no good way to measure this other than manually digging today
B) My experience has been that while it may matter, you can certainly rank quite well even with links that come from either very broad topic sites or sites that have nothing to do with your site at all. Think of links to places like Flickr, Twitter, widgets from finance sites that are consumer-targeted, etc. Maybe it helps, but it certainly doesn't preclude a link from passing value.
BTW - we are looking more at topic classification things in our newer ranking models. That work should be done by Q3 of this year.
Thanks a lot for your reponse, and good to hear you're working on thematic classifications. You may well be right, surely link bombs and your "click here" and "leave" examples show that links work pretty well without being thematically relevant.
Still, permit me to remain the devil's advocate for a bit more. I have thought about this some more and realized what it is what worries me in various discussions of links is the phrase "trust / authority" (a lot of people seem to have picked up this equation of trust with authrority). First of all, to my mind trust and authority are two different things. Trust = confidence that a site is not spammy; it is not theme-specific. Authority is theme specific: an authority on pancakes does not necessarily give the most authoritative weather forecasts. I think these should be separate metrics. What troubles me further is both somehow propagate via links nd bear some relationship, not entirely clear, with good old PageRank, which we associate with link juice. It seems to me increasingly that some of the "buckets" are partially overlapping and some of this languagelanguage may bear some further refining in the long run.
Anyway, from this angle, what do you think of the possibility that relevanceis not merely icing on the cake but an integral part of Authority? It seems intuitively hard to accept that Google would discount relevance since search relevanceis what Google is fundamentally all about. Why would Google not use it?! I am fully open-minded about this but can't help wondering.
I think Rand addressed pretty well...
It isn't so much as relevancy doesn't matter, but that it is still just one aspect, and if the links are coming from relatively good sites to begin with, they add value. Relevancy may just be a little extra icing on the cake, but to even determine how much icing is quite challenging in an environment that is in constant movement.
Assuming link building as an ongoing process, then we might also assume that the overall link profile is going to naturally skew to being more relevant anyway, just through our own efforts and those links that are being accumulated naturally from others linking to our content.
In other words, probably don't need to be overly concerned....at least today.
It's those correlations that are the tough part. The sites with extremely low relevance are also usually low-authority sites with tons of links that may be operating exchanges or selling links. Teasing it all apart is crazy hard.
Hi, Dr. Pete! Agreed, it's very hard to measure this, but hey, isn't that what SEOmoz is for?! :) Here's how one might go about measuring something like this, I think. So say a certain page ranks at some position for a certain SEO term, for example. We give that page a followed link from some highly relevant SEOmoz page and measure the change in ranking when it stabilizes. We then, by some arrangement, give the same page a link from a freshwater aquariums (or whatever) page of the same value (as best can be ascertained according to the other 4 "buckets") while removing the SEOmoz link, and measure the ranking change. Repeat wiht different linking and landing pages [insert you favorite large number] times to make sure. The more powerful the links, the more visible the impact and the differences will be.
I wonder if relevance ties into "neighborhood" somehow. When I see 3 suspicious links, for example, from a low-quality directory - 1 to an online casino, 1 to a Viagra distributor, and 1 to a small business site (all in the same div), I can't believe that doesn't impact quality. The question is, what exactly is Google measuring in this situation?
I think many people hear "relevance" and assume semantic analysis - that Google is somehow comparing the anchor text of the link (or the target site) to the content of the page/site where the link lives. I think that's pretty unlikely at this point. On the other hand, Google might see that 5 links all end up landing in 5 very different site clusters or neighborhoods and determine that the overall relevance is questionable.
Just to help me understand better: do you mean that - following your logic - if I have a natural link from a website that in the same page have other links to other sites which have a bad neighbour, that means that my link is getting affected?
Are you talking about trustrank somehow?
TrustRank is a Yahoo affair, no relation to Google.
I think semantic analysis is too strong. If Google doesn't need semantic analysis to determine a page's relevance to a single term, it doesn't need semantic analysis to determine the mutual relevance of two pages either. Is I may speculate a bit, semantics is not strictly necessary to determine thematic correlations, statistics do just fine! In Google Webmaster Tools Google shows tells you what terms it thinks are most relevant to your site. There's no semantic analysis involved. I bet Google can do the same for an individual page too, and go from there.
How many unknown tools are still developed? I would not want to work on the same issues, if you allready build one I need. Because no tools, only OSE for a linkbuilder, but it has no aI layer over it.
This is a great WBF
My question is two fold
should you diversify your anchor text?
Is it detrimental to not diversify your anchor text?
When I use OSE I often find little varitation in the anchor text and wonder if this is beneficial.
Shawn,
I understand that anchor text should be at least rotated with the keywords you are working on, because the anchor text used is the keywords the SE's will rank you for (which is why Disney ranks for "leave" and "exit", like Rand said).
Not changing anchor text can be a signal to SE's that the links are unnatural.
Thank you Chad,
I have been cycling keywords in the anchor text but what cuirous since using OSE I saw most sites used the same anchor text.
Rand--
You had mentioned in this WBF that I need to take a look at our competitors' links and see how they've done in their "four buckets." In the interest of wisely-spent time, how deep should I go into this? The first ten pages of results? Would that return a pretty good indication of their four buckets? Or should I go with something like twenty or thirty results? Or more?
:)
Honestly, I'd just do the analysis on the top 10 and your own site. There's barely any visibility/traffic to be had past page 1 so comparing yourself to page 2-50 is likely a lot of wasted time/energy.
Okay thanks! Will do.
:)
Great whiteboard Friday! I did have one question however based off the link-authority issue in big buckets:
In a particularly a non-tech-savy market, people love linking to a product producers website. The non-tech-savvy people all have really low quality domains, and end up usually linking to the product makers website. The product maker has several really good links, but the majority of links are definately coming from these low quality domains in the limited market, in this case would those be hurting the company?
Alternatively I can ask if needed in the Q&A just let me know =)
I don't see why it should hurt the company. Technically is that a form of link baiting??
As long as most of the smaller sites aren't involved with black hat seo I really don't see it hurting the company's ranking. It might not make the company sky rocket to number one but it'll push it in the right direction.
Thats what I think anyway... :)
Thanks Tola, good to hear - I was mainly concered as the market is a small market and most sites look circa 1998 in their look and structure - and have few to no inbound links.
Also read your post the other day :P Good work.
Very helpfull. Especially when arguing with your superior about those topics, who says "just buy some links. I know a guy who can get us to postion #1"
One thing that is very difficult to explain to new clients is that different links are worth more than other links and why having a certain type of link from certain types of websites is also more valuable to your website.
Far too many times have I seen the navigation on a splash page be what the page is ranked for, it's ridiculous.
Anyway, thanks for this WBF.
great video on highlighting the importance of anchor text.
thanks
Great pie chart (4 buckets) - that's great to laid out like that!
Also love the idea of finding what metrics are causing your compeitors' links to work for them and then going and using that data for your own link building. Or even finding what your missing with your links. I think a lot of people go half bind into link building and this posts gives them some light on the subject.
It would really cool to see a tool that laid out what types of links a site has and makes recomendations on what to do from there. Maybe I'll put something together like this... :)
great explanation of the basic principals and the importance of quality over quantity.
Does the attribute 'name' or 'title' within the link have any SEO weight or value?
If it is an image enclosed the link the title of the anchor helps. If you have a text link you could test, by adding a little more relevance in the title attribute. For the second i am not sure, how much it improve, i never have make an extensive testing, my strategy about link building is never used the same behaviour, i am always doing so much variations as possibly, so i have used title and in the most cases not, do appear more casual.
Maybe our SeoMoz team has done a research about this?
Time to open another excel spreadsheet!
With all these 'new' tools that seem to get mentioned on the various blog posts (all of which seem pretty cool and useful), does any Mozzer have a suggestion on how to persuade management to pay for Pro??
Thanks for the post Rand.
That's hysterical about Disney ranking for exit and leave!
They don't for me when I checked, maybe they contacted Google & they removed them...
I'll continue to look on this site but what I really need to know (can you point me in the right direction?) is the results of a study or research (not opinion) on which type of link (say from the same site so that all other variables area equal) passes the most juice:
a contextual link with the "right" anchor text within a post created simultaneosuly with the post
a contextual link with the "right" anchor text within a post, created AFTER the post was created
a comment link with the "right" anchor text
a text link with the "right" anchor text that is not either of the above (e.g. footer, blogroll, etc)
just found the answer to the old vs new contextual links here
https://www.seomoz.org/blog/new-links-old-pages-whiteboard-friday
Simply awesome!
I can't believe I just watched a massively old video. Great help though; I'm currently on a crash course of link building trying to absorb as much info as possible.
Finally got around to watching this and honestly it has been excellent. It has cleared up why some article sites and social media sites have pages ranking so highly.
I've seen plenty of admittedly good articles on the squidoos and ezinearticles etc rank for fairly competitive terms with what appear to be low quality backlinks. Now I can see the combination of their own domain authority and a diversity of low quality backlinks (with some good anchor texts in the mix) to those pages is doing the trick!
Looking forward to the new tool!
Thanks for an other great WBF.
I understand that inbound links is important. The biggest question is how to get them for a small site on a limited budget and without getting on a "link farm" website. So many areas in SEO seem very time consuming and link building can be the biggest sucker of time. I read somewhere (perhaps on this site) that something like 1 out of 300 link requests will get a contact and perhaps a link back. Outside from the theory of why - what works to get quality links?
I suggest you to chek out this post by Rand, especially to check out the third Scribd presentation: "The Key to Rankings" by Eric Enge.
My suggestion? Creativity and intelligent networking. Remember that links are finally reccomendations toward something that is considered valuable... so, what can be valuable right now for other webmasters or people tweeting and sharing links (the tweeterati, a sort of mix of Tweets and Linkerati) about the topic of your small site?
Try to ask that question and you will be walking in the right direction.
Thanks a lot of this clear explanation on how to analyse what can be wrong with my netlinking strategy. Now, as Fjonan said, it's even easier to explain it to superiors and convince them.
I agree with the tool suggestion, even though (as you Rand demonstrated with practical cases in the last two webinars and probably in the next one on May 11th) the combination of OSE for a general perspective and Keyword Difficulty Tools for specific keywords is a powerful way to study how these four factors are balanced or not in your links' portfolio.
And I think that this method can also be applied to new websites with few/no links still. Infact, adding these analysis to the study of your competitors can help you planning your link building strategy.
As for the 4 factors, I believe that the anchor text is probably the most important one, but I think that it's also needed to talk about anchor text diversity.
Infact, and this is can be considered the other side of the "click here" coin, many times I see websites that have backlinks all with one or maximum two optimized anchor text. And - here I'd like to have a positive or negative feedback - this uniformity can be somehow dangerous as it can look unnatural to the Google's eyes.
I hear what you're saying, but I've actually not seen the "consistent anchor text" hurt people so much as the source. Look at the example I gave in the video - Posterous has literally thousands of unique domains all linking with the exact same anchor text. I doubt those are hurting them at all.
What I worry about is when folks go out and manually link build or link buy from low quality sources and use the same anchor text. If you're doing black/gray hat link acquisition, A) stop - you're better off doing other things and B) don't make your trail even easier to follow.
Thanks Rand for the feedback.
And... yes, I was mainly thinking about the second case you talk about in your answer.
Now I go to check out the Posterous example
Do you think it makes a difference whether or not the anchor text is the site's domain/brand name (as in the case of Posterous) or a competitive generic search term (like "credit cards")? The former feels natural; the latter, if replicated identically thousands of times in cookie cutter fashion, seems to suggest manipulation.
So, this raises an interesting question. How do you think Google determines (and penalizes, in some cases) badge and widget links?
It would seem that there are only 3 quality factors at play, in theory: (1) the inbound links come from sites with low-relevance, (2) the inbound links all have the same anchor text, (3) the links build up too rapidly at too high volume. Discussions on this post alone reveal the ongoing debate about whether (1) and (2) matter (and how much). What do you think the determining factor boils down to?
Similarly, I've been seeing (anecdotally) links from clients get devalued. Take, for example, website designers who slap their "Built by Company X" link on every site. This seems like it would have to be either an anchor text or relevance issue, in my mind.
I don't really think relevance, similarity or growth rate trigger penalties on their own. I've seen plenty of sites/pages that have links go widespread on low relevancy pages with high growth rate and exactly the same anchor text (widgets, badges, embeds and a lot of linkbait have exactly these characteristics).
I think the engines really care about the sources of those links and any patterns that make it look like the link/embed was not editorially selected. Rough sampling can make that pretty easy to spot, IMO and Google likely has a huge repertoire of spam characteristics they can pattern match against.
Great WBF. I think wakeless is right on with a tool that incorporates all of this would be awesome.Rand....go to bed.
I finally earned enough trust to gain access to streaming video from work. Whiteboard Friday was my first destination. Great presentation, Rand. I'll get working on my links ASAP!
If your competitor is ranking better than you for some longer tail keywords, 4 words or more and I do not believe it is because of anchor text links, which one of the other three could be the factor? Our trust is very similar. The only difference really is they have 3 times more domains linking to them (and linking pages), but most of that is probably images linking to them from ramdon sites.
Having links from random sites help in situations like this, increase your own site's trust & authority up enough, your own internal SEO could play a big enough role to get you ranked without any anchor text links for those specific long tail searches?
I'm not sure if this makes sense...
Quite a revalation that Disney ranks for the search term exit.
All of a sudden it makes clear why some search results on long tail keywords are so bizzare. They must have anchor text links distorting their search results.
Brilliant as usual, Rand!
Great WBF as always Rand. Look forward to any new tools for link metrics.
Great WBF! Thanks!
I'm wondering... In fact, are sitewide links a bad thing? I mean, even if you plan on having a link on some of your partner website, is it better to have it not sitewide?
I also had a lot of fun with the allusion to Disney vs p*orn websites. I remember Crayola was being used a lot for these websites too...
Well... It's not that I know!! Me? I never visited this kind of website...!!! :)
Personally I don't usually ask for a link to my clients.
But those times I ask it as a way to endorse their success also thanks to my collaboration, I ask to put just in the homepage footer as a credit.
I know footer links are not the best ones as they tend to be devaluated, but footer still is the less obstusive place where to put a credit link (or using the tecnique that SEOmoz use in this site... just check the link to distilled and exact target here below in the footer).
Double check those links - they actually point to pages here on SEOmoz that talk about why they exist and then link to Distilled/ExactTarget. They're technically not classic "sitewide footers" which we do think the engines have been devaluing in many cases.
Exactly, that was what I was trying to say. To use the SEOmoz tecnique in order to avoid the footer "penalty" and give weight to a more editorial credit page, that finally have an editorial link out to the credited company website.
I mean, it's a really good tactic that - if done honestly and with real editorial purpose - can be used almost for every site.
Hi,
this is a good tip. Would you say that this technique is recommended for all type of sidewide links, indepent from their location?
I have for example more than 1000 backlinks from a content relevant forum pointing from each page at the right hand area the same link and anchor text to a client side.
Could it be more effective to try that the forum people make me a partner page and then pointing one or two links to my client site?
Thank you for this valuable information.
Yes, my problem for my main job is that my main competitor at SERP 1 (i am at 3) has round about 1000 Link Rooting Domains more.
I have not the bugdet to rent links from 100 Domain, etc.
So i was thinking to use a submission tool for webdirectories worldwide to get this factor better, but this is probably spam because my client has german content.
Greetings,