News Source: http://www.seomoz.org
Published: Mar 17, 2008 - 11:41 am
Story Found By: Wiep 1891 Days ago
Category: SEO
14 Comments
Sorry this site requires JavaScript to be enabled in your browser. See the following guide on How to enable JavaScript in Internet Explorer, Netscape, Firefox and Safari. Alternatively you may be blocking JavaScript with an advert-related or developer plugin. Please check your browser plugins.
News Source: http://www.seomoz.org
Published: Mar 17, 2008 - 11:41 am
Story Found By: Wiep 1891 Days ago
Category: SEO
14 Comments
Upcoming Conferences
Search Engine Land produces SMX, the Search Marketing Expo conference series. SMX events deliver the most comprehensive educational and networking experiences - whether you're just starting in search marketing or you're a seasoned expert.
Join us at an upcoming SMX event:
- SMX Paris - June 6-7 2013
- SMX Advanced - June 11-12, 2013
- SMX East - October 1-3, 2013
- SMX Stockholm - October 14-15, 2013
- SMX Milan - November 7-8, 2013
- SMX Social Media Marketing - November 20-21, 2013
- SMX Israel - January 2014
- SMX West - March 11-13, 2014
- SMX Munich - March 25-26, 2014
- SMX Sydney - April 2014
- SMX London - May 2014
Latest News From Marketing Land
- Facebook Dominates Teen Social Networking But Enthusiasm Waning Over Drama, Stress & Oversharing
- Pew: 94% Of Teenagers Use Facebook, Have 425 Facebook Friends, But Twitter & Instagram Adoption Way Up
- Marketing Day: May 21, 2013
- Nutella Gives In, Allows Fan To Continue With “World Nutella Day”
- How To Organize Your Informational Content: A Case Study
- - See more news at Marketing Land -
Upcoming Webcasts
Learn more about search marketing with our free online webcasts and webinars from our sister site, Digital Marketing Depot. Upcoming online events include:
- Taking your PLA Campaigns to the Next Level - Jun 4
- Increase Shares, Tweets, Likes, and Pins to Fuel Search and Drive Engagement - May 30
- Take a Closer Look at Links – Why a Site Ranks - May 21
- The Future of Digital Marketing Services: Where Will Agencies be in 2014? - May 15
- Unlock the Key to Successful Forecasting and Conversions with Big Data - Jun 20
Research:
- B2B Marketing Automation Tools 2013: The Buyer’s Guide
- Enterprise SEO Tools 2013: A Buyer’s Guide
- Enterprise Social Media Management Software
Whitepapers:
- PPC Bid Management Guide
- The Marketer’s Guide to Driving ROI From Facebook Advertising
- Collaborative Customer Interactions Management
- How To Create a Data-Driven Dynasty
- Top 10 Ways to Improve Digital Experiences
Latest News From Search Engine Land
Comments
a) This was already proven last October:http://sphinn.com/story/8887b) He’s giving bad info in the post and comments. Currently you cannot sculpt PageRank from one link to another link of the same target using nofollow. Check any of the phrases used in the anchor texts of the links in the test in this page, adding nofollow to the first one didn’t do anything:Single Source Page Link Test Using Multiple Links With Varying Anchor Text - Part Two
Michael, perhaps it’s your b) link that we should all be sphinning. It makes sense, which may or may not mean that Google actually always does it.
bwelford, thanks. I never did write a follow up post explaining what the results meant with that test, perhaps I should.
I’d certainly be interested in hearing it :) - it would be an interesting topic for discussion
The idea of using nofollow to "shift" which link’s anchor text counts is nice, but I don’t think it actually works. Relying on that kind of undocumented and speculative behavior, even if it worked once, can cause some real headaches down the road.If you want to scuplt PageRank with nofollow, all links from page A to page B should be nofollowed. You either cut the link between the two pages or you don’t.@Michael, thanks, I thought I’d seen something published a couple years back, maybe longer, but certainly that Sphinn showed up well before this one.
"The order of operations is dedupe first (remove all duplicate links), and then apply the nofollows."Interesting.
Whatever SEO impact the current behavior has is interesting in the short term, but it’s important to bear in mind that whatever they’re doing is designed for efficiency, and subject to change for any reason. We have older tests that give a different result, so it’s likely that this behavior has changed at least once.
"it’s important to bear in mind that whatever they’re doing is designed for efficiency, and subject to change for any reason."Agreed Dan. I’m not sure if this behavior will change though. Say you got a several links on a page pointing to the same target URL. First (top of source) is an image link with anchor text "home", second is a contextual link in an article with anchor text "las vegas", and third, a footer link with anchor text "las vegas real estate." Two ways to handle this: all anchor texts apply OR only one applies (according to this article, the current set up). Now, ideally Google would choose the "best" anchor text ("las vegas real estate") out of the three, but that would require some pretty intricate logic - coding that into Googlebot would probably really slow it down.
Now, ideally Google would choose the "best" anchor text ("las vegas real estate") out of the three, but that would require some pretty intricate logic - coding that into Googlebot would probably really slow it down.Halfdeck, I couldn’t swear to this 100%, but I’m fairly confident that actual anchor text analysis is not done by the bot, but rather at the ranking algo end. It wouldn’t actually slow down Googlebot.
"I’m fairly confident that actual anchor text analysis is not done by the bot, but rather at the ranking algo end."Yeah, that’s also the way I look at it mvandemar. My point was if Google wanted to handle things differently, it may need a more complicated Googlebot. I guess Google can record links on a page during crawl, then handle dupes post-crawl, but that would still require more number crunching than the current set up, which is more of a *hack* than anything else. It’s straight-forward to code a bot to choose the first link, but if I had to code a scraper to choose the "best" link, I’m not sure how I’d go about coding that.
Right. Assuming you need to de-dupe the links because you don’t want parallel edges on your graph, then you have a couple choices. You can mash all the duplicate links together into one big link, or you can pick one.If I were writing the algorithm, mashing them together would make more sense, because you aren’t losing the anchor text when bloggers say "Pat Johnson(link) wrote an amazingly stupid thing about widget margins(link)." If you just take the first link, you lose the topic of the target URL.All these tests are showing that they pick one link to use, and apparently it’s the first one. Any notion of efficiency would go out the window if you had to run some elaborate routine on every page to pick the "best" link, so that first-link-wins behavior does make sense.You see all the time where people nofollow the "Home" link that appears at the top of every page, and add a footer link to the home page with their keywords stuffed in. The problem is that you may end up not getting either link followed - you’re relying on Google to remove the first link and still follow the second link. Highly speculative.
"I guess Google can record links on a page during crawl, then handle dupes post-crawl, but that would still require more number crunching than the current set up, which is more of a *hack* than anything else."Actually, I’m was thinking all they would have to do is switch the order of operations around... first remove the nofollows, then remove the dupes. That way there would be no conflict and it shouldn’t affect performance at all.
I wonder whether 301ing fake landing pages to homepage would allow multiple links with different anchor text pointing to the same URL ? Or some non-existing parameteres ? That would be an additional thing for the algo to calculate and easy to circumvent, but who knows?
"Actually, I’m was thinking all they would have to do is switch the order of operations around... first remove the nofollows, then remove the dupes. That way there would be no conflict and it shouldn’t affect performance at all."Good point, though that way they still would end up with the first followed link, not the "best" link.