Some Clarity on Paid Links

Hoo boy, with that title, this better be good …

The paid links discussion raged hot and heavy at Search Engine Strategies San Jose last week. Whenever the debate comes up again, there is a lot of confusion and misinformation that gets put out there. Keeping in mind (here is the ritual disclaimer) that while I do not work for Google, and do not represent Google, this post will outline just a few components of how I believe this works.

What the Issue Is

Google’s founders, Sergey Brin, and Larry Page, had a thesis (a Stanford University thesis to be exact) that you could measure the quality of a web page by the number of links it receives. More precisely, some links would count more than others, as sites that had more links pointing to them would have more “link juice” to vote for other sites, and that link juice would be voted on a mathematical basis (if you linked to 100 other pages, this means each page gets 1/100 votes of your page rank).

To work in a reasonable manner, this algorithm assumes that the people providing the links are not aware of the algorithm.

This is the essential point that we all need to understand. The base algorithm works best if people don’t try to manipulate it. It’s just like what happens with tests that show how rats find cheese in a maze. Once the rat figures out it’s being observed, the behavior changes dramatically.

This is the algorithm which launched a company that currently has a $158B market cap, and which is the most important search engine on the planet. For purposes of clarity, Microsoft has a market cap of $262B, and they are scared to death of Google, because it is currently dominating the market space that represents the future.

As soon as a link is compensated (paid for or bartered for) this model starts to break down. We can demand that Google come up with another model, but replacing this model is just not simple. You can bet that Google is working on alternative or supplemental algorithms as hard as they can. For example, the recent moves towards personalization are one step in this direction.

Ultimately, relevance is king for Google (and the other search engines). This is the fundamental fact of their business. Higher relevance scores equates to changes in market share, and is what drives Google’s algorithms, and how they choose to modify them. At this point in time, a model that relies heavily (but not solely) on link analysis is still the best model, even with its flaws.

Of course, there are people who buy links for the purposes of increasing branding and traffic. Simply put, this is not wrong. I will go further than this, and say that people buying links for purposes of influencing their search engine rankings is not wrong, either. These people are simply trying to win at a game where the rules are defined by Google.

Google has stated that they are not out to punish people who buy links for innocent purposes, and that they have no issue with people who buy links for branding and traffic reasons. However, Google will try to detect these links, and discount them, because it does not fit into the model of counting links as editorial quality votes.

Ultimately, we need to understand that Google, and the other search engines, will process the links they find as they see fit, as a part of determining which web site ranks higher than another in their search engines. We can’t fault them for that, even though they make mistakes from time to time. This is their business. Best relevance wins.

Some Examples

In a recent post on link buying on Search Engine Watch, I wrote about what I believe the Google policy is. The first thing to realize is that Google is not saying you shouldn’t buy links. They just don’t want people to do it for the purpose of influencing search engine rankings. They will do whatever they can to keep their algorithm in balance, by trying to detect paid links and remove them from the equation. If they succeeded in removing all paid links from the equation, then the original logic of the Brin-Page thesis still works and applies.

With this in mind, let’s look a bit deeper at some observations about link buying:

  1. PPC ads sold by the major search engines are not considered paid links. They have never conveyed any link juice at any time, and never will. They are very easily detected by the search engines, and are ignored for organic ranking purposes.
  2. Buying a listing in the Yahoo! directory is not considered purchasing a link. The reason why it’s not considered a purchased link is because of the following:
    1. Yahoo! actually does not sell links. They sell an editorial review service. What you are pay for is to have a human editor review the listing.
    2. They reserve the right to change your proposed listing, if they think your categorization, and/or description are not optimal.
    3. They reserve the right to reject your listing altogether, and note that Yahoo! does reject many proposed listings.
    4. If they do reject your listing, they keep the money you paid. In other words, once you have paid, they keep the money no matter what.
  3. The cumulative result of all these things is that a Yahoo! directory link still gets placed based on editorial judgement that Google respects. To put this in other words, Google has found that crediting Yahoo! directory links is something that helps increase their algorithm’s overall relevance. Another example of a directory that I believe passes link juice is Best of the Web. There are probably many directories that pass link juice, but beware, there are far more directories that don’t pass any link juice at all.

  4. Paying a third party firm to work on obtaining links for you is not buying links. This presumes that they don’t turn around and pay web sites to link to you. There are many link building services that perform a PR like function of going out and spreading the message about your site, and getting people to link to it simply because they like the site.
  5. Getting links by syndicating quality content is not purchasing a link. This is because the site taking the content is in fact endorsing your content by taking it. There are other issues to watch here though, such as that of creating duplicate content on the web. At STC, when we push out syndicated content we send out only original content that is not on our site (or our client’s site) to help deal with that issue.
  6. Swapped links are likely to be treated as compensated links. This is one of those gray areas where I am sure there are lots of exceptions to the rule, but straight up link swaps are barter, not an editorial endorsement.
  7. The standard response to finding a paid link is to discount it, so it passes no link juice.
  8. The way to become exposed to more severe penalties is to engage in deception. If Google sees deceptive practices at work, such as a clear case of manipulative link buying, then they may penalize the site. Ultimately, really severe cases can result in banning. You don’t want to go there, because once you have engaged in deception, it’s going to be hard to persuade anyone to let you back in, even after you stop engaging in the deceptive practices. Simply put, your credibility is shot.

Summary

The biggest gray areas occur when people buy links for traffic and branding purposes (although I believe that it’s a minority of people who buy links solely for this reason), and in the handling of the various types of reciprocal links. The gray area for swapped links comes in when the link provided to your site was in fact an editorial endorsement, and your link back was also in fact an editorial endorsement. I would think both of those links should pass link juice, but it’s a scenario that is hard to detect.

The lack of clarity on what works, and what doesn’t is frustrating, but I don’t think that this will get clarified in the near future. Too much clarity on these policies makes the job of people who engage in deceptive practices much, much easier. The best we can hope for is that Google (and the other engines) will come up with supplemental algorithms that make the whole thing less of an issue.

As this unfolds, expect the emphasis to grow on sites with deep content quality and a superior user experience. Ultimately, this is where all the search engines want us to invest our time.

Comments

  1. “To work in a reasonable manner, this algorithm assumes that the people providing the links are not aware of the algorithm.”

    The algorithm assumes more than that. It assumes that people will only link because they believe in the quality of the destination, but that was never the case. Even in 1998 people were linking for pay, for the sake of “completeness”, because they were engaged in communal linking arrangements, because they were being asked to link back to content providers, and for other reasons that had nothing to do with the quality of the destination.

    The PageRank algorithm has nothing to do with determining relevance, as Brin and Page’s original paper explained. The relevance algorithm was used to score results which were then weighted with PageRank.

    PageRank is now used as a quality filter on the other end — Google divides the Web into pages with sufficient PageRank (the elite pages admitted into the full Main Web Index) and pages with too little PageRank (the pages — probably the majority of Web documents — that are stored in the Supplemental Results Index).

    But Google also says it uses PageRank to help determine crawling priorities. In a natural crawling system PageRank would only be an indicator of the probability of a crawl, but Matt Cutts and other Google representatives have made it sound (to me) like Google actually favors pages which meet some internal PageRank threshold requirement.

    PageRank is also used as a reward system. Google allows pages to pass PageRank if they comply with its Webmaster Guidelines, and they allow pages to accumulate PageRank if they comply with its Webmaster Guidelines. Hence, Google has perverted the original PageRank concept from a probabilty measurement to an credibility valuation.

    That transition from probability to credibility has forced Google to become intransigent about paid links and other link manipulation schemes. Most people who have manipulated links professionally were doing so primarily for the anchor text (which directly influenced the relevance scoring). But now Google has given Webmasters compelling reasons to pursue PageRank solely for the sake of PageRank.

    PageRank was never very important to RANKINGS, but now it has become extremely critical to INCLUSION.

  2. Thanks for the mention Eric – I couldn’t have said it better myself. Both the Yahoo and Best of the Web directories are prime examples of using a high-quality editorial review process to ensure directory quality and relevance. And you said it best here – “To put this in other words, Google has found that crediting Yahoo! directory links is something that helps increase their algorithm’s overall relevance.”

    Great article – thanks for revisiting a very important but often misunderstood fundamental concept of search engine marketing!

Speak Your Mind

*

*