We've been hearing some requests lately for some really advanced, expert-level content, and this post is here to deliver. I've built up a short list of topics that deal with more cutting edge SEO, and if there's interest in this series, I'll try to make it a regular part of the blog. These tactics aren't black or gray hat (we're not advocates of that kind of thing), but they're very specific in use and tend to be at the opposite end of the "low-hanging fruit" basket.
The first in the series touches on a common SEO problem - determing if a link has value and how much. This tactic isn't low effort, so it should only be employed when the link or link source is particularly critical.
Testing Whether a Page/Site Passes Link Juice (and How Much)
Scenario: You've found some potentially valuable, but possibly suspect link sources. These could include things like a seemingly high quality directory that requires payment or a site you're worried may have aroused Google's ire for one reason or another. The need for a credible answer applies anytime you're unsure whether a link is counting in Google's rankings and need to know.
Tactic: Find a page that's already in Google's index and a somewhat random combination of words/phrases from that page's title and body for which it ranks in position #3-10. For example, with the query - https://www.google.com/search?q=new+york+presentation+morning+entitled+link, my blog post from last week on Link Magnets ranks #3. The query itself is not particularly competitive and the pages outranking it don't have the exact text in the title or domain name (a critical part of the process).
If I now place a link with the exact anchor text from another page (like the blog post you're reading now), e.g. new york presentation morning entitled link, I should be able to see, once this post is indexed by Google's spider, whether it passes link juice. The result will be positive if the page moves up 2-4 positions in ranking and I can be fairly assured that the link is indeed "Google-friendly." With that knowledge secure, I can change the anchor text and/or repoint the link to the desired location. I don't simply use the anchor text I want initially because with competitive queries, a single link may not make enough difference for the ranking impact to be visible and I don't want to waste my time/money/energy.
(Metrics displayed in the SERPs via mozbar)
Special Requirements: To make the testing work, you'll need to be able to repoint the link, change the anchor text or 301 redirect the linked-to page (though the last of these is the least desirable, since 301s lose some link juice in the process and good anchor text is so valuable for ranking in Google). Also, here at SEOmoz, we don't recommend buying links, so while this tactic could be applied to that process, remember that manipulative links may later be devalued, wasting all that time and effort you spent acquiring them.
Results: With this technique, you can not only get a yes/no answer to questions about whether the link passes ranking value, but a rough sense for how much (depending on the position change - this can be a good reason to use pages that rank in the #7-10 range). Do take care to record the ranking positions of all the pages in the results and leave the test running for 1-2 weeks (longer if there's very fresh results ranking for the query). If you don't, other factors may conflate to hide the true results.
I'm looking foward to your feedback about this technique - and let us know if you're interested in seeing more of this advanced/edge-case content on the blog, too. Below, I've listed the topics I could tackle in future "Advanced" level posts.
- Hosting Pages on Third-Party Sites
- oDesk/Mechanical Turk for Content Development (and Link Research)
- Email Marketing for Search Personalization
- Modifying Product/Business Naming Conventions
- Spiking Search Volume and Capitalizing on QDF
- Protecting Inter-Network Links & Domain Acquisitions from Devaluation
p.s. If you do like this kind of thing, I'd also suggest:
- Register for SMX Advanced: Seattle or SMX Advanced: London - both are quite good and SEOmoz will be sending speakers to both. You can use the code SEOmoz@SMX for a 10% discount to either event.
- Check out the SEOmoz Expert Training Series DVD, which just launched last week. The video alone will get you pretty excited :-)
- PRO members should check out our libraries of tips, video content and webinars.
I'm in Tampa, then Miami this week, but will finally return to Seattle for some much needed time in the office next Monday. Until then, blogging, commenting & email may be a bit slow from me.
Google also gives juice to links that are NOT indexed
Links need to be crawled by Google (use google alerts) and you can get value.
Most ppl think that the page needs to be
1. indexed
2. have a page rank
for the backlink on that page to have any value ...which is false.
I think I'd disagree witht the first part. A link that has not been indexed can't pass juice. If Google hasn't seen a link, there's no way for them to count it.
On the PageRank front, I totally agree - plenty of pages that don't yet have PR assigned via the toolbar are still passing juice.
but if you think about it what you wrote it is incorrect
Google doesn't index every page they crawl
Just because they crawl a page doesn't mean it will be index
Just because they crawl a page doesn't mean THEY DON'T count the backlink on that page
"If Google hasn't seen a link, there's no way for them to count it."
Google sees a link by CRAWLING the page not INDEXING it
Sometimes they crawl a page a few times before they index it
Yeah, that makes perfect sense to me. If only indexed pages passed link juice, setting the meta robots on a given page to "noindex, follow" would be completely pointless.
Indexed or nonindexed pages pass link juice
Setting meta tag to "noindex, follow" is another flip of the coin that I haven't tested or tried
I still think I'd disagree. A page that Google has chosen not keep in the index is likely not one they're going to collect and run link calculations upon. If you consider the Order of Operations to calculate things like PageRank, Trust, etc. it requires the index of pages being built up first. I also think lack of indexation is a pretty clear signal that the engine doesn't want to count links from that page.
do some social bookmarking dofollow linkbuilding (just 1 example)
a lot of those pages don't get indexed but rankings go up
i have multiple case studies to show the results
if i have the time to write it, i'll submit the article to seomoz
it's probably one of the best linking strategies i know
1. it makes my rankings go higher
2. my competitors can't see my backlinks
Agreed, except in cases where the page has been noindex,follow'd -- this technique is often used for dupe content pages or low-value pages from a content perspective but high-value from a crawl point, link equity perspective.
by the way that was for Rand not you Robert :)
Rand - I need some clarification on this one.
For example, we have a series of sitemap pages that provide very little value to users (just a bunch of links). We have them to encourage indexation and pass juice through to our deep landing pages. We were considering noindex-follow until we read this post. We decided not to based on your advice that noindex pages can't pass juice.
Then I read this interview with Cutts where he directly confirms that noindex pages can still accumulate pagerank and pass juice.
https://www.stonetemple.com/articles/interview-matt-cutts.shtml
You actually commented on the interview on Sphinn? https://sphinn.com/story/8667
Am I missing something on this one?
I agree with Robert. I had a page on my site: no content, no incomming internal links, not indexed (noindex, follow), but TBPR 1 (1 external link). I think that page must pass PR.
i agree with you
great post. SEOMOZ needs more advanced articles like this one. Vote for "Hosting Pages on Third-Party Sites"
Amazing information as always as well in this article, thank you!
oooh! I second the "Hosting Pages on Third-Party Sites" :)
"A link that has not been indexed can't pass juice. If Google hasn't seen a link, there's no way for them to count it." - That comment confuses two different areas, being crawled and being in the index. I think the earlier idea that a crawled page may offer some power through its links (though perhaps minimal) is worthy of testing, but I can't see a method for that right now. Nevertheless, it brings up the question of how Google builds their webgraph for PR calculation. If I were Google I might well include many more pages in the webgraph than I surface in the public SERPs.
It's certainly an interesting concept. I'm guessing that based on how the link graph is built, how calculations are run and how they store and process multiple indices, non-indexed but crawled pages really are "dropped" and don't have their links, anchor text, page and domain metrics, etc. sitting around in memory somewhere waiting to be part of a link graph calculation, but it's not impossible.
Also - great to see you here, Ted!
Rand, further to our Q&A, I will take full credit for the idea of this post. Thanks :)
Credit where credit is due! :-) Thanks again
You had me at "I'll try to make it a regular part of the blog."
One thing people need to keep in mind though is that this test is also specific to the tested phrase, especially in regards to the "how much" portion.
So one it tests out and the ranking bump is 4-6 positions, when the anchor text is realigned to the "real" phrase, which is no doubt more competitive, the same amount of lift may not be achieved...perhaps it moves up 2 positions, or perhaps none, until a few more links are also acquired.
Doesn't diminish the value of the test, just need to recognize that when the equation is changed, the result may change as well.
Interesting. The only problem I see is getting in a directory that will let you change your link after a few days.
I did like the article and would not mind seeing similar ones in the future.
Yes, performing this test on 3rd party sites may be quite challenging.
But it certainly doesn't limit the testing within your own site. Iterate this test across your own site's pages to help identify the stronger pages and their potential.
Nice post - I'd certainly vote for more advanced content but then I'm biased :-) (also I have to help write it!!! d'oh).
Is there anything you can do to check if existing links are contributing to rankings? That would be almost more valuable I think for competitive research...
Right. The scenario where you can not only get the link, but then change the anchor text after that is few and far between in the mentioned example of testing whether or not a directory has been spanked or submitting to some other site that has been disciplined in the past.
Doing a search for a multi-word snippet that is common across several directories or websites and seeing if the suspect site is omitted from the SERPs is still a viable method for gauging a site's quality it seems.
Adding an intitle:some-word to the search (using a word from the suspect site's page title) in addition to the snippet that is present on the suspect site should really confirm as it shouldn't be listed in those results either.
Hmm... I'm not sure I've got any solution to the question of "does this link I already have and can't modify pass value (and how much)?"
There's just so many ways that data can be conflated. :( I'm certainly open to ideas.
I don't know if "how much" is as important as "is this positive or worthless/poison?"
Your method does open up a new business model: link escrow services.
Only agree to pay a directory after the link has been proven to pass value using a unique identifier targeted at a temporary, but indexed page hosted by a third party escrow service. Once it passes the test the link is made to the actual page with the appropriate text, and the directory is paid. The longer the history of successful tests and pleased customers the better the reputation of that directory.
Pie in the sky: Going the Linkscape route you could have Nick try and come up with Linkscorch, an index that approximates the supplemental indexes of search oblivion... It's a bit fanciful for how HUGE of an index that'd be, but it'd be pretty impressive to see a simple Mr. Yuck sticker in the toolbar indicating the current page is likely poison.
On the toolbar side of life, try to identify on-page problems that my cumulatively indicate spam to the search engines: hidden text, no contact information, masked links, etc. could sorta kinda help.
If optimization is a tug of war between good and bad choices, most tools out there are great at identifying how to pull harder for good, but there are few that succinctly tell you how to lessen the pull of the bad. I liked this post as it went a little further down that less bad route. Work both sides of that battle and the work is that much more efficient.
Of course the short answer: will be try to get links from real people. :)
First of all... where are you going to celebrate Easter? On a plane?
Second, I will follow your tip, especially in order to make a cleaning the directory listings, as in Italy and Spain the directory panorama is quite confused with sites which really give value and other that could be considered suspicious.
About your list of "Advanced" post, here my ranking:
1) Modifying Product/Business Naming Convention... as it's something I deal with quite often working with clients eCommerce;
2) Protecting Inter-Network Links & Domain Acquisitions from Devaluation, as many clients have build a network of products sites, microsites, related sites with optimized domain names;
3) oDesk... because I've only 2 hands and 24h per day (and a family too to enjoy);
4) Email Marketing for Search Personalization (even if I can imagine where do you want to go talking about);
5) Hosting Pages on Third-Party Sites, as it's a tactic that more and more I'm suggesting (as guest blogging);
6) Spiking Search Volume and Capitalizing on QDF
Other advanced post (or so it should be for me) could be more on a "pr/vendor" perspective: How to convince a stubborn client that his beloved multimedia-ajax-flashed website is not going to make him happy. I know it's something we have talked about so many times, but a recap and opinion exchange (now that you have left the consultancy field) would be surely important.
Ciao!
Great info! I would LOVE to see more posts like this. I'm sure that my visit frequency would increase if you covered more advanced topics more often. Maybe you could add an icon of some sort at the top of each blog post noting if it's beginner level, intermediate or advanced. It would also be sweet if I could follow a specific RSS feed associated with each level.
more advanced content would be welcome :)
I really like stuff thats tried and tested, and working in the current SEO climate so to speak
It's a good, but slow strategy to find out if a page transfers Pagerank, but the 'how much' part isn't that realistic. Besides that, it's hard to explain and control the results of such a study, because you a) do not know all variables that changes rankings, b) cannot control all variables. For most SEO's this practice would take to much time.
If you test it with an external link to a non-related 3rd party website, it might not have the same effect when you change the link. Since external links are also part of the ranking algorithm, it can have diverse effects for both pages in terms of ranking, also depending on the relevancy between the pages.
Diverse, not adverse right? Maile taught us all about Spammy Links in an excellent video SEOmoz has out there.
I LOVE THIS POST!
My vote, for what it is worth, is more technical impactful content like this in your blogs. PS I followed this from FaceBook. Had it not been for your Fan Page, even though I am a registered member, I would not have seen it.
Kudos SEOmoz for bringing great content to the WWW again.
TrafficJam,LLC
WoW! some creative thinking has been cooking here! This is a very good idea, but not all that easy when today webmasters are so lazy that adding the first link is like asking to rub your tummy and pat your head at the same time. Have to click "Add Comment" before I can try...
Really nice tactic and informative post Rand, thanks! I will go and test to see how this works for us.
I think some more advanced post on QDF would really be helpfull, but then again, which topic isn't?
I'd love to see more advanced content here - my vote would go with the oDesk/Mechanical Turk for Content Development/Link Research.
I am a little late, but am I missing something...
Doesnt this technique require you to be able discard the link if you find it unhelpful? Like a money back guarantee.
So you would have to be able to convince the link source - hey I want to test your link to see how powerful it is, if I find it unsatisfactory than our exchange back to you is off.
Where do you see this technique actually being used? Maybe in the negotiations of a complex partnership between 2 sites. I dont see it working for any typical directories. Or maybe if your site already has links from a source that you can manipulate the url and anchor text, you could move this to a testing page on your site and make sure it is positive - if yes you could then go back and determine what page on your site you want to improve - if is has very low impact maybe you link it to a deeper page on your site rather than the homepage.
Voting for mturk. Although if it gets too popular, the rates may go up too.
OT, Congratulations on achieving PR8 for seomoz home page!
Good information as always!
Please please please do write more advanced content. The basics are sometimes interesting too, but there aren't many places to get advanced SEO.
It's need much time to judge whether a Page passes link juice..but it still worth to have a try.
I want to do some Email marketing, so i want to know your opinion about Email Marketing for Search Personalization.
thanks for sharing!
Great advanced post, thanks! Whilst finding third party sites where this was possible probably be a problem though, I think the benefits of being able to try it out on others is really worth it
Great post - I'd definitely like to see more of this type of content in the future!
My votes are for:
1) Protecting Inter-Network Links & Domain Acquisitions from Devaluation
and
2) oDesk/Mechanical Turk for Content Development (and Link Research)
I am still eating up lots of the information on SEOMOZ to help my clients and myself. Still, even tho compared to many of my esteemed peers who frequently comment and avidly follow *randfish* and his crew of wild and crazy teckies, I am a SEO fool, STILL, I find the advanced tips forcing me to stretch my brain beyond the bounds of decency and leap into an optimized unknown. Beyond the rhetoric: DO IT. My vote: Yes for deep sea diving for Randfish.
Thank you Rand! Please do try to make these sorts of things a regular part of the blog. I definitely appreciate the need for a heavy focus on the beginner/intermediate posts but it would be great to see a few more advanced posts each month.
Rand, How do you test the value that internal pages gives?
I vote for Spiking Search Volume and Capitalizing on QDF
I believe you could use this same principle on internal links (and it will be even easier).
Great advanced material Rand. The more like this the better.
Andy made a good point above. How much of an effect does the freshness aspect play a part I wonder.
As far as the advanced topics you mentioned, my vote is: Modifying Product/Business Naming Conventions
As someone new to SEO, I would like to learn as much advanced stuff as possbile. I actually end up understanding other things on a much deeper level by reading the more advanced stuff. Thanks again Rand for helping me choose between SEOmoz and that 'other' site.
Great article! tackled something we're dealing with right now - A competitor has used one of our brand names to rank above us. They are listed first in the SERPS as page title= "Our Brandname" blurb= "If you're using "our company's" "our product and brand" consider making a change to us..." While our page is better targeted towards the brand name keyword they're outranking due to a higher page rank. I'm now trying to get as many links out there as possible with "our brandname" keyword as anchor text. This link juice testing process should help.
In the future I'd like to hear about Hosting Pages on Third-Party Sites
Thanks
Client's competitors messing with brand names is a pet hate of mine. Had to push one off page one completely recently.
Cover yourself on PPC. Then try to dominate local search so you can take up more real estate for your brand while you build the links.
You can also play with strong domains such as LinkedIn to build brand profiles that rank.. get the best internal links you can from employee profiles, blogging, etc. eHow profiles (created for your brand) used to work too but they are having issues at the moment.
Best of luck Daniel.
appreciate the recommendations! We're shelling out for the PPC but would much rather reserve that budget for non branded keyword terms that are generally more difficult to rank for. Linking from Employee profiles is not a tactic I'd considered but a clever one. Thanks!
Just recommend they join LinkedIn - you get the link to the LinkedIn Company profile when they put in their job title. Depending on the number of employees, how many connections they have, etc. you can get the LinkedIn company profile #3-5 for your brand term by doing very little. Extra bonus if they use My website to link back to your site :)
Agree that LinkedIn is important, though links from individual profiles may not necessarily pass PageRank any more.
I wrote about this on Natural Search Blog last December, LinkedIn, But NoFollow Link Love. Some interesting observations in the comments as well about some links not having nofollow applied (like My Website, My Blog).
But links should always be about more than just PageRank, so even if all of these links become nofollowed some day, still important to be there.
Youtube and Vimeo are good options as well as you may even be able to get more bump from universal results.
As a precaution you should secure your brand name across as many of the "most popular" sites across the net: twitter, facebook, myspace, etc. so that the name is secure there and you have the option of building out those profiles in the future.
If your name really is branded (i.e. copyrighted, trademarked, etc.) there are legal options that you can pursue as well.
You can use https://knowem.com/ to check and quickly secure your brand name across multiple platforms. Most have profile pages with a link back to your site which can help, some are follows most are nofollow
I need to come back again to made reference to this post. I read all but is still blank in me.
Perhaps I'm too naive to absorb what's being taught.
Anyway thanks you in advance.
cheers
Very useful method for testing link quality.
I am afraid there are quite some places where it isn't possible to run these tests (or very cumbersome) but it makes a lot of sense to do it from sitesyou can get links from on a regular basis, e.g. blogs where you guest post.
My vote is on Mechanical Turk for Content Development. I have noticed you talk alot about Turk, and I think it's interesting to use it for SEO. But I'd like some more details - the suspense is killing me ;)
I love this advice, thanks!
What about links that you cannot control, does the formula TBPR/#ofLinksOnPage give a good ballpark estimate? Some advice around estimating value of links on pages/site you do not have editing rights would be great!
Thanks again!
*edited for misspell
Not necessarily unfortunately. We've seen plenty of pages with TBPR, low numbers of links, and other good metrics fail to pass link value.
This post is #1 in the UK for that phrase already rather than that other URL :)
Vote from me for "Protecting Inter-Network Links & Domain Acquisitions from Devaluation"
Hmmm - not only is this post at No1, lots of other results from sites that evidently publish the SEOMOZ rss feed.
I like the theory, but I do think you need to factor in some time (possibly a week) - with Google's current love of "fresh" results, you need this post to slide away before you can really guage the result. Right now it's at No1 because it's fresh.The linked to page isn't that old yet either.
Good point Andy. QDF might also be a factor.
Yeah - I'm definitely thinking QDF is at work here.
Awesome. Thanks Rand!
I love testing and will certainly be trying this out.
My vote definitely goes for "oDesk/Mechanical Turk for Content Development (and Link Research)". I would be REALLY interested in reading what the mozzers has to say on this! :)
I'm still wondering how to test this principle using websites other than directories... Ppl in the SEO industry keep on telling us that most of the good links are the links we attract; I don't see myself telling to this or this blog owner "Hey... Change your backlink text for XYZ please".
I know you already gave us this technique requirements Rand, and I'm sure it's efficient... I'm just wondering if it's also feasible. And in my case (Small, foreign business in the web design industry), the anwser would be "no" for 99.9% of my backlinks, unfortunately.
However, thanks for the great information, as always. I'll be looking forward to use this technique in a "feasible" way, on future backlinks! ;)
Well, therein lies the rub. When you're naturally attracting links, my guess is that 99.99% of the time, you don't need to worry about whether they pass juice (they probably do, but you haven't invested much effort in it anyway) :-)
Well... It's so obvious it's almost embarassing! Forget what I said ok?
Guess sometimes we try to dig things too much...
*Owned by Rand* :)
How about we just go autopilot with this, you guys come up with a cool new graphic to add to the pro member tools, and then we get a link quality evaluator tool :)
Ha ha! I think we're working towards this with metrics like Page Authority, but we certainly need to spend more effort on perceiving/identifying spam in the same ways Google does before we can really get there.
Awsome rand... I'll try and use this on my suspected links ;)that's great!
sharp idea - thanks for sharing!
i'd also definitely be interested in more posts along these lines - esp. 'Hosting Pages on Third-Party Sites'
interested in seeing more of this advanced/edge-case content on the blog!
Awsome rand... I'll try and use this on my suspected links ;)
Great post Rand, I've been hoping for some advanced technique posts.
As far as the voting goes, my practical Left Brain says "oDesk/Mechanical Turk for Content Development (and Link Research)" while my creative right brain likes "Modifying Product/Business Naming Conventions"
also I think I need to take a break from staring at monitors because when I glanced at "Spiking Search Volume" I thought it said "Spiking Beach Volleyball"
very informative posting, thanks