Moderator: Detlev Johnson
Intro: The one thing that you can do with your own site is content and text to help gain rankings, but link building takes a slightly different approach. Can have a tremendous effect, example of PageRank that gave rise to people persuing links �obsessively.� Link buidling can be legitimate and work, but it can also be less than so. Unfortunately Eric Ward is caught in Atlanta and will not be here.
Mike Grehan - Smart Interactive Ltd. Feels that UK is one step behind US in SEO and Canada is one step behind UK. Shares a little perspective from the meeting he has a few weeks ago in San Jose with the Yahoo engineers regarding the perspective from the outside looking-in about SEO. One of things very important is to understand how the SE�s and their crawlers rank documents. Created a great schematic that is now being used by SE�s to describe the process. Gave a short history of ranking algorithms. In the beginning Brian Pinkerton created WebCrawler and designed a method of ranking. When a SE builds an inverted index, it first analyses all the words that appear on the pages by indexing each one. Vector space model, term weight pair: a page has a weighted list of terms, and a term has a weighted list of pages. This meant that rank could be manipulated by using a term more often. Thus the rise of SE spamming. This caused Brian to fade into the background.
Then John Kleinberg tackled the abundance problem and �bestowed upon us the hub and the authority.� Google takes the idea and wraps it with Markov Chain to create PageRank, and make it possible for spammers to get on the top by manipulating links. Asks now when people want to rank first: why should you be? Tells everyone that under one chair there is a sticky note with that question and he wants that person to stand up and say why should their page rank first. Turns out he was kidding - he likes to see the look of terror on people�s faces. (laughs)
Hyperlink analysis makes two assumptions: Link from A to B is a vote for the other. Second assumption is that these are related topics. Co-citation helps to determine classification and categorization. C to A and C to B, but A and B do not link. Even though they do not link together, there is a connection. In order to reverse engineer an algorithm you have to have this kind of data, which is owned by Google and others. Hubs and authorities show that some links are worth more than others (Side note from Mike: many people suffer from GAS at SES conferences, they need help�GAS stands for Google Anxiety Syndrome, what did you think he was talking about? Laughs)
Cyber communities: which do you belong to? You do not want links from bad communities or links that are not from your community. An example would be 100,000 communities from 40 million sites in japan.
Top ten important parts about linking: 1. Quality not quantity 2. Anchor text in links 3. SE�s have already decided who is important, so you can use their info to get linking partners. 4. Don�t dilute quality content by trying to spread it all over the site if not needed. Create an authoritative page that people will want to link to. This increases links to that page based on different reasons/content that others want to link-to on the same page. 5. Affiliate program can bring lots of traffic and sales, but can cause havoc in linking strategy. 6. Do be choosey about whom to link to. �If it�s free, it�s probably crap.� 7. Faking linkage data by creating 5000 links from your own fake domains. This will work initially, but then SE�s will see that no links go to those. 8. Asking for a link is a business process like anything else. 9. Acquiring links is time consuming, and the use of software doesn�t necessarily guarantee that the links will be �good.� Needs to be personalized. 10. Instead of trying to be Yahoo and create a directory, you should concentrate on creation of good content.
Debra O�Neil-Mastaler - Alliance Link
Linking reminds her of the Overtstock.com ad that says �it�s all about the �O.�� In this case �it�s all about the �A�,� which means anchor text.
Linking Myths and Mistakes Reciprocal linking is bad - False. Allows you to place targeted kw anchor text links on pages/sites of your choice. However, you need to have diversified links and not use ONLY reciprocal links. Need to stay in community. Don�t use e-mail templates to gain them, personalize! Once again, quality over quantity. Check to make sure link is on the site you agreed to exchange with and not on a 3rd party site. Be sure to use anchor text. Don�t email sites not already swapping. Add a �how to swap links� page to help new partners. Think outside the box when looking for link partners. Great success: agrees to swap ad space in newsletters or blogs; the best part about this is that it becomes archived. Links that age are thought to be more valuable.
Non -reciprocal out-bounds are bad- False. Engines analyze sites by their inbound and outbound links. Helps establish you site as an authority or hub. No site is an island, only inbound may raise flags. This WILL NOT �leak PageRank.� Try to link to a site that mentions you.
DMOZ greater weight - Unlikely, but it is a very credible link.
PR 3 or above links only count - False. Google knows about all of them, but will not necessarily show them. Google simply chooses to not display all inbound links. Nothing more than that.
Incoming links from bad sites will hurt- False. You cannot be held accountable by who links to you. Try to secure links from authority sites using targeted anchor text.
Need lots of links to rank well - False
Debra then talks about Eric�s presentation: Links come in all shapes ands sizes. , just like websites. A quality links can mean different things for different sites. Shows szapos.com, not much text, but still ranks consistently Number one. They pay for links, get links in blogs (that they found because someone mentioned them and went to them to ask for a link), press releases, editorial links, passively obtained organic links.
If you wanted to promote something short-term, use things like discussion groups (mostly a powerful traffic driver, some do not have links that count), private fan links, Imdb Page that reporters use. Mix links to make sure they don�t all come from the same IP block, and not all from .edu�s or .gov�s, for example.(Erics� presentation and examples can be found at : http://netpost.com/ses/link_building_basics.html) Best to be well=rounded in anything in life, link building is no different.
Q&A
�Are blogs more spider able than forums, or less? Is one more helpful than others?�
M: Not necessarily, it depends how the forum or blog is structured. Also in blogs and forums there is sometimes not enough text to (contextually) qualify the link.
Debra: Agrees, the traffic is probably the most important part of getting a link in a popular forum or blog.
Detlev: Some use the
�Are link farms bad?� M: Yes if a site just give away links to anyone from anyone in any community.
D: Check for links inbound to site, and use PageRank as an indicator of popularity. There�s only about 35-40 good �general directories� and the rest are niche.
Detlev: You are not going to get credit from a link that is not indexed, use a search to see if the directory/site is indexed first. Teoma.com does break out authorities and hubs nicely. Look for �links from resources and enthusiasts� to find hubs.
�How effective are software tools to automatically gain links?� Are they useful?
Detlev: Can they get you kicked-off? Google even spells out do not use webpositiongold to run links, because they do not have the authority.
D: agrees: if they do not have an API, you can have problems and get banned due to automated queries.
M: If it brought in 5,000 candidates while you were asleep, you still have to go through them manually.
�If a site doesn�t have PageRank, but has a high Alexa, should it be considered?�
M: Let me tell you something: Google DOES NOT use PageRank, it�s a myth. It�s green fairy dust.
