Search Engine Q&A on Links

Mar 1, 2006 • 3:32 pm | comments (2) by twitter | Filed Under Search Engine Strategies 2006 New York
 

Moderated by Detlev Johnson – Position Technologies. This is a question and answer format session, and four search engines are represented on the panel. Aaron D’Souza from Google, Kashual Kurapati from Ask.com (formerly Ask Jeeves), Rajat Mukherje from Yahoo, and Ramez Naam from MSN Search.

Kashual starts with a short presentation from Ask.com. The general basis of Ask’s method results in the greater the number of links, and the higher quality of the links pointing to a page, the more “authority” it has on the web. Ask was previously Teoma, they use “subject specific popularity.” 1. Search the web and index info. 2. Break the index into communities. 3. Collect and calculate local subject specific information and bestows “hub” and “authority” status on particular sites. 4. Once query comes in, the results are ordered from more to less relevant, applying all pertinent global and local information found.

“More links is better” isn’t always the case. If you gather “spammy” links, you may not get much benefit from them. If you are going to purchase links, be careful. It’s like buying recommendations…how can the search engines be sure to trust these recommendations? Be cautious of: reciprocal links and purchase links. Avoid: link farms, cloaking pages, invisible ort hidden links that try to trick the crawler. Generally they advise that you should become an expert on your subject. If you focus on your business and content, the rest will follow. They use the subject specific method to develop “ExpertRank,” one of their algos.

Next up is Ramez Naam from MSN. “Why do links matter?” 1. They are how SE’s discover pages. 2. Links can help figure out the popularity of pages. 3. The links are good descriptions of the pages (the text within the links themselves help). The basic principles of link building, according to MSN: 1. Build links that real users will click on. Don’t use 3 point font on white background, etc. Think about what links are natural. 2. Build/create good content, and links will come organically. These links will matter most.

Five specific tips for good links: 1. Make them descriptive (don’t use “click here”). 2. Put the links in the main part of the page. Links “hidden” will get less clicks, so they will be considered less valuable by the SE. 3. Keep URL’s short and readable. Crawlers may have problems. 4. Beware sessions ID’s and query parameters. Ie: different URL’s for every unique visitor can make it difficult for the spider. 5. Point to the pages that you actually want to have show up in the search results. Avoid all links going to home page.

Think twice before doing the following: 1. Paid links. These are usually created to fool the SE’s, and they’ll figure it out. 2. Link exchange pages…not user friendly. 3. Link farms, usually a easy to find “bad link.” 4. Don’t use blog and forum SPAM. Remember: you may fool us for a little while, but a technique that works today could get you into hot water tomorrow as the SE’s improve their ability to detect “bad links.”

Rajat from Yahoo is next. He says “my real name is Tim Mayer.” Then after the laughs he says that Tim is busy with work and a baby boy and apologizes that he couldn’t be here. Just wants to mention that hubs and authorities and using links to rank results was actually invented by John Kleinberg when with IBM (CLEVER), and not any search engine. Since that point, the SE’s have incorporated this and other factors in helping to rank pages. In the case of buying links, he says that you need to remember that authoritative links are more important. Speaks briefly about Site Explorer, a Yahoo product that helps you identify a variety of things, including backlinks (in-links) indexed by Yahoo. This interface is meant for webmasters and publishers, and offers many features. Encourages those in attendance to play with this tool and give feedback. Then shows a bunch of links in order to contact Yahoo, including ysearchblog.com, and the one to check ona site: help.yahoo.com/search/sitereview. Also help.yahoo.com/search/reportspam to report spam, and to send feedback (not support): use ystfeedback@yahoo.com.

Aaron from Google is last, and he says instead of making a formal presentation, he just says “what they said.” (laughs)

Q&A

Q: Part of the rank is allegedly determined by the text near a link as well as within the actual link. Is this the case, and would a site with a certain focus have an advantage since it may have a linguistic consistency that is better than another site? Ramez says that the consistent use of particular words in linking to you will help in your rankings (huh? I thought we needed to vary inbound anchor text…) Detlev ads that CNN has a site that speaks about a large variety of topics, but a link from them is considered good. (for some reason I feel that maybe the other reps may have been asked not to comment too specifically on this excellent question)

Q: “We have millions of pages of content that we feel is valuable. We exposed it in a directory structure to make it easier to find... Wants to know how many links to place on each page, or to go many levels deep with fewer links? Rajat: there are time constraints that limit the number of pages/links that will be deemed relevant and actually crawled or return-crawled. Kashual…what is better for the users? Agrees with Rajat that there is a limited time to crawl a site, so who knows how much will be crawled and not crawled. Aaron ads that it is difficult to determine what pages the webmasters really want indexed. They (G) will use more tools like sitemaps to help webmasters actually specify what pages should be indexed (coming soon).

Q: What about rich media content that can’t be HTML or text, like art sites, etc…? Rajat: Yahoo has provided ways/feeds to get content indexed, once having gone through an editorial context. Kashual: Ask now supports Flash, and they are trying to catch up and get the info indexed. Aaron says it is still difficult due to arbitrary binary formats that differ greatly. There is more easy-to index content such as PodCasts and other “tagged” information. Evolving standards for rich media content are always being developed. Rajat says that the specific audio and video search type engines are making things easier. Aaron ads that you should make sure content is visible to software for disabilities that “reads” pictures. Detlev echoes that without any sort of tags, it is difficult. Providing the means for people with out plugins to get the content will also help the SE’s. Rajat says that enterprise search vendors have focused on these problems more.

Q: If I have a keyword ranking before others, will that help once competitors start using those keywords? Aaron: you would hope that if you used it first, most of the anchor text with that term will be helping you to be the “authority.” When you do misuse any given signal, at the end of the day, “fairly stupid” computers are analyzing the info and will eventually diminish the value of links, since they may not necessarily be a reliable method of determining relevancy any more. Kashaul ads that sometimes “the loudest may drown you out.”

Q: Internal linking structure, using “home” on the top navigation may not be useful.. Can you use something in the footer that is more descriptive? Second question…what about reciprocal links and they seem to still work even though he has heard many bad things. Rajat: in general, we treat internal links different from external links, but that they can still look for some “gaming” going on with internal links. Ramez: if you have 3 or 4 partners, for example, that you usually do business with and have lots of “deep links” to various page, this will be fine. But if you have a reciprocal links page that includes all kinds of off-topic links, it won’t help. If they are just there to accrue more links, there will probably not be a long term value to them. Kashual uses the SEM community as an example…they are used to the various links to SEO’s and SEM’s, but if suddenly “gambling” links appear, there is obviously a problem. Aaron again using SEW as an example…you would imagine that SEW has the majority of its links coming from non-reciprocal sources. In this case a few reciprocals won’t “hurt.” Essentially, you shouldn’t use too many. Rajat ads that in general there are innocuous users and egregious users. Only egregious users will see a negative effect.

Q: How does algo detect difference between a good hub and a spam farm? Kashual: a good hub also gets linked-to from relevant pages and links out to relevant pages. However, if in a “bad neighborhood,” you will get problems by association. Aaron says think about a “rats nest of links” like Wikipedia, which is valuable but very different from a link farm, which IS just a rat’s nest. Regardless of the status of a hub or not, the content of the page is important as well. Ramez reminds that they cannot divulge exact information, but remember that the system is getting better, so as it does more methods to fool the system are caught and dealt with.

Q: Everyone says don’t buy links, but even Yahoo sells “express inclusion” into its directory. What determines which are legitimate? Also, can you give a percentage of how important link building is on a whole compared to content development, for example? (some jokes on the panel about 5.3%, 27.9% etc…laughs) Detlev says it’s important to remember that this emphasis is a moving target, so what is right today might not be right tomorrow. Rajat says the good news for Yahoo is that they send project managers that don’t know the answers to the percentage questions to the conference on purpose (laughs). The point about the directory: getting into the directory doesn’t necessarily mean you’ll get in the results. We can in some sense trust some directories more because of the approval process. Aaron: he has heard the question of paid links= advertising or not? They are equivalent…but who is the target? A paid link targets SE’s and an ad targets users, which is the big difference. Ramez: the answer on how important links are depends on the query. If you search a brand, chances are the one with more links is the actual brand. However, a long tail search might return a result with just a few links.

Q: Is there a limitation to how many outbound links should be on a site? Ramez…if a page has thousands of outbound links, we know that users may not be able to see many of them. Others did not comment on this…:(

This is part of the Search Engine Roundtable Blog coverage of the New York Search Engine Strategies Conference and Expo 2006. For other SES topics covered, please visit the Roundtable SES NYC 2006 category archives.

SES NYC Tag:

Previous story: In House Forum
 

Comments:

ScottW

03/02/2006 06:17 am

Great post Chris! I wasn't able to make this SES and was looking forward to this session. You summed it up nicely. Care to go back to grad school with me?

Megan

08/31/2006 06:48 am

Just wants to mention that hubs and authorities and using links to rank results was actually invented by John Kleinberg when with IBM (CLEVER), and not any search engine.

blog comments powered by Disqus