Interview by Fredrick Marckini
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

AltaVista (AV) is an undisputed veteran of the search engine industry. Their initial search engine endeavors trace back to the mid '90s. AltaVista has made use of their early lead to remain a forceful competitor in this ever-changing industry. With over 500 million Web pages indexed and more than 40 million search queries a day, AltaVista remains a major force in the search industry. I had an opportunity to catch up with AltaVista's Erica Aks, Manager of Content Engineering and distinguished Engineer Edgar Whipple. The topics of our conversation ranged from the best ways to submit, to the importance of knowing who your virtual neighbors are. This is an interview search engine optimizers should read with great interest. The team we interviewed is AltaVista's first line of defense against spam.

Partners
~~~~~~~~

AltaVista has strategically teamed with other industry leaders in order to enhance their service offering. Currently, two of AV's main partners are GoTo and LookSmart. AV uses GoTo's paid listings in their search results as "sponsor listings." Aks and Whipple noted that the GoTo listings have really worked well in conjunction with their own results and AltaVista plans to continue serving the GoTo listings into the future. AV also has close ties with LookSmart, the popular search directory. Right now, LookSmart supplies all of AltaVista's directory content, and AV plans to continue this collaboration in the future.

Search Results
~~~~~~~~~~~~~~

Although AltaVista noticeably separates their directory content and their Web search section, this does not necessarily mean that directory listings will not appear in the AltaVista Web search results. Erica noted: "There are a couple different ways that you can actually access directory information. You can go directly to it from the front page, you can take a normal Web result and then click on the directory tab and get just directories. And then we also index the directory content, serve it up as well within our normal Web results but we treat it with the same relevancy algorithm."

I also learned that AV does not call attention to the directory results that are mixed-in with the Web search results. There is no highlight or pointer indicating that a particular result is from LookSmart's directory. Apparently, a listing within LookSmart's directory is exceptionally important for success with AltaVista! Search engine submitters, if you want to maximize the visibility of your Web site, particular effort should be given to obtaining a listing within LookSmart.

Related Sites - How this works
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Edgar stated that the selection process of the Related Sites feature is similar to the classic information retrieval concept of co-citation. He went into detail to explain how this works: "If I write a paper in the academic field and then publish it and then people start citing my paper in their papers, then A) that's a good thing because it means that I've written a good paper but what will happen is they will cite presumably other papers than mine when they're writing their paper so they will cite my paper and perhaps somebody else's paper at some other university. So co-citation says that if paper X cites my paper then the other papers that it cites along with mine are likely to be related."

Submitting Web Pages
~~~~~~~~~~~~~~~~~~~~

AltaVista understands the shared value proposition between AltaVista and the Web site submitter in indexing unique content. AltaVista advises all submitters to follow a basic rule in deciding which pages to submit to their engine. "For indexing a Web site, the general rule is as long as all of the pages are unique or largely unique and they have quality content, we want as many of them as we can get," noted Erika.

Erika also commented that the AV indexing policy adjusts sporadically. The level and depth of how far the spider will crawl a particular Web site varies from time to time. Consequently, if you have pages that you genuinely want indexed, submit them, since the spider may or may not find them otherwise! In addition, if you have a hierarchy of directories at your site, put the most important information high, not deep. AltaVista assumes that information placed higher is more important.

In recent months, AltaVista modified their Web site submission process so that it is no longer possible to submit Web pages through an automated process. This was discussed in the April edition of the MarketPosition newsletter: "AltaVista now requires you to read a graphical submission code off their page and input it into their form. The code purposely uses difficult to read font styles, presumably to prevent anyone from using OCR technology to read the codes automatically. This has effectively blocked automated submission products and services from submitting URLs to AltaVista." Web site submitters, keep in mind that automated or programmed submission tools will no longer work with this engine. As most MarketPosition readers already know, you must now submit each page you would like indexed by hand.

Gaining Rankings within AltaVista
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Frames
~~~~~~

If you have information inside frames, you probably face a challenge gaining rankings in AltaVista, but it is not an absolute barrier. AltaVista indexes the outside of the frame as a distinct page. It will also index each pane of the frame window as a separate page. That means that if the content matching a query is in a pane, then visitors clicking on those links will see only the pane, not the full page as it was originally designed. So if you want visitors from AltaVista to experience your pages in a certain way, you should have non-frames as well as frames versions of those pages, and submit the non-frames versions.

META Tags
~~~~~~~~~

AltaVista has a firm view of META data, keyword repetition and comment tags, which is discussed in more detail on their site at "Being Well Indexed." AV declares "basically, META tags are a band aid to help you deal with pages that don't state what they are about in clear text, right up front. Do it right to begin with, and you don't need META tags at all. Many webmasters think that by using the keyword META tag, they are gaining some advantage in the ranking or making up for the fact that their pages have very little text content. But, according to AV, those words are worth little more than any other word in the main text of the page: "There is nothing 'key' about it. You have simply added a few more words to the page in a place that is not visible."

This statement clearly suggests that AltaVista places little value to the Webmaster assigned META data section of a site and therefore assigns low-weight to META data in their ranking algorithm.

Keyword Repetition
~~~~~~~~~~~~~~~~~~

Be very cautious of how many times you choose to repeat your keywords either on the page or in the meta tags. AltaVista does not reward Web pages that practice useless repetition. AltaVista only counts each unique word twice.

Comment Tags
~~~~~~~~~~~~

Comments, that is, text between symbols in the source code, are not indexed at all. AltaVista feels that they are intended as private communications, not viewable by Web site visitors, except by using View/Page Source and as such are not considered by AltaVista's algorithm.

Link Popularity
~~~~~~~~~~~~~~~

AltaVista considers the links pointing at your site from other sites across the Web in determining relevancy. However, the text describing the link (anchor text) is very, very important in deciding if the link is pertinent. I asked my interviewees if it was possible to be negatively affected by a link pointing to a site. For instance, if, for malevolent purposes, someone points a link to a site indicating that the site is a great source for adult material, when, in actuality the site sells greeting cards, will this negatively affect the site's ranking? I was informed that if the anchor text in the link pointing to your site does not appear on your Web page, the link will not be considered and therefore will not harm a site's ranking or relevancy within AV.

Consider that AV is so concerned with the anchor text contained in the link pointing at your site that they validate the destination text before they reward the site for the link. It emphasizes the importance that AV places in the content of the anchor text of the links that point to your Web site. When you're out soliciting links to your Web site to improve your rankings in places like AV and Google, remember to ask not just for a link but specifically request that the link contains one or more of your targeted keywords or phrases in the anchor text.

Dynamic Content
~~~~~~~~~~~~~~~

AltaVista recognizes that for both malicious and unintentional purposes, some Web users may create dynamic content that causes robots to get stuck in an infinite trap of page generation. As a result, AV typically does not crawl dynamically generated sites. However, if someone submits a dynamic URL through the manual submission process, it has an opportunity to be indexed in the database. Submitters, if your Web page has great content but contains a query string, go ahead and submit it by hand as it still may be indexed. Dynamic content is not an absolute roadblock in submitting to AltaVista.

Refreshing the Database
~~~~~~~~~~~~~~~~~~~~~~~

Both of my interviewees agreed that the latency between spider visits to sites within the AltaVista database is growing shorter and shorter. In order to compete with the exponential growth of the Web, AltaVista has enhanced their technological capabilities to spider on a more frequent basis. While no specific waiting time was indicated that a site should expect a spider to revisit, it was noted that AltaVista makes every effort to ensure that they revisit "the most popular pages the most frequently." We infer that they're referring to link popularity here.

Know Your Neighbors!
~~~~~~~~~~~~~~~~~~~~

The concept of virtual hosting, where a Web hosting company may host 10 or 12 Web sites on the same IP address, was a revealing topic of discussion during our interview. Erika acknowledged, "What we see on the Internet is that there are the IP equivalents of good neighborhoods and bad neighborhoods. So we see it fairly frequently where people on a certain IP exhibit a certain kind of behavior..."

Therefore both iProspect and AltaVista recommend that if you opt to participate in virtual hosting, be aware of who and what types of other sites are being hosted on your IP address. If other sites who share your hosting IP address are engaging in spamdexing, either submitting too frequently, optimizing for irrelevant keywords, etc., AltaVista may block several or even all of the Web sites hosted on that IP address.

If you've ever encountered problems gaining rankings in AltaVista, even after you've properly optimized your Web site, you may want to think about moving to another hosting provider or if all else fails, starting a new domain name and changing hosting providers at the same time. We've encountered rankings problems that we can only attribute to what you might call, a "dirty domain" problem. As the interview revealed, AltaVista (and probably other search engines) may be banning sites hosted on certain IP addresses and IP addresses "near" yours, e.g., numerically close to your IP address.

By changing your hosting ISP, you are doing more than merely asking your ISP to change your site's underlying IP address. The goal of changing hosting providers is to change to a different class-c subnet, not merely a different IP address on the same class-c subnet. Here's a description of the different components of your site's IP address:

255.255.255.XXX here the X's identify the host of a class-C subnet
255.255.XXX.XXX here the X's identify the host of a class-B subnet
255.XXX.XXX.XXX here the X's identify the host of a class-A subnet

The goal of changing hosting providers is to change your class-C subnet. If your current hosting provider has enough class-C subnets, they may be able to move your site without you having to change hosting companies. In most cases, changing your class-C subnet is sufficient. If search engines were to block class-B or class-A subnets they'd be blocking a big part of the world as these designations identify so many potential users!

For whatever reason, sometimes a "fresh face" to the search engines makes all the difference. The AltaVista disclosure above is just one of many domain and IP related issues that could interfere with your site's ability to obtain top rankings. In our many years as a search engine positioning firm, iProspect has identified several dozen hosting, domain and IP address related issues that have interfered with a Web site achieving rankings in search engines. Changing hosting providers and/or the domain name often solved the problem and allowed the site to gain strong search engine visibility when all other efforts failed.

Many versions of Windows allow you to open a DOS window and type PING followed by your domain name to determine the IP address of your Web site. If you're unsure of your IP address and whether other sites share it, contact your hosting service.

Time Matters
~~~~~~~~~~~~

Are you not finding your Web page currently for a particular search query? Yet, it came up in the search results for the exact same query just three hours ago? Don't worry. My interview disclosed that queries made during different times of the day might yield different results. Consequently, if you conduct a search during a peak traffic time, it is possible that you may receive different search results. Using load balancing, AltaVista distributes the site activity evenly across multiple servers in order to prevent a single server from being overwhelmed. While your ranking may exist on one particular server, there is a chance it may not be present on another.

Fighting Spam
~~~~~~~~~~~~~

AltaVista considers themselves to be the best in the industry at fighting what they consider spam. Erika and Edgar noted several ways that a site could get dropped from the AltaVista search engine. Their list included: hidden text, tiny text, excessive keywords, duplication of content, pages consisting only of links to other pages, Web pages that only exist to redirect users to another page, and so on. In addition, don't try and sneak in duplicate content, AltaVista has an exclusive patented mechanism for detecting such Web pages.

Although there are countless activities that AltaVista deems to be unethical that risk being dropped from the index, AltaVista again provides a general rule to follow. Any action taken by a promoter "that would not create a positive user experience" could be labeled as spam and could cause a site to be blocked.

Subsequent to my interview, AltaVista performed their fourth site redesign in just over a year. Here are some of the changes that have taken place:

• There are no longer tabs encouraging users to try AltaVista's vertical search options.

• The previous "Related Searches" links can still be found at the top of the page with the introductory text "Others searched for" coming before the links.

• Numbers next to results have been dropped in favor of bullets.

• GoTo.com sponsored listings are now labeled as "Partner Listings" though listings sold by AltaVista's ad department (or by doubleclick) are still labeled as "Featured Listings."

• Page Clustering has been altered so that you may see up to two pages from the same web site. The second page appears "indented" below the first page.

• Vertical search links now appear at the bottom of the page, rather than in tabs at the top. These let you run your search against AltaVista image shopping, multimedia and LookSmart directory databases.
  

Evident in the recent adjustments made, AltaVista is constantly fine-tuning their search services. My enlightening conversation with Edgar Whipples and Erika Aks of AltaVista reinforced my belief that AltaVista is, and will remain, one of the sought after search engines. As the Web expands and search technology progresses, I am both excited and anxious to see what AltaVista has in store.

                 ← What is this?

---