Pagerank Should NOT be Gotten Rid Of

In case you’re wondering, it’s been 5 months since the Google PageRank was last updated.  Before the flames start, I should clarify that statement.  PageRank actually updates dynamically.  What is normally called a PageRank “Update” is really just the pushing of a snapshot of the dynamic PageRank out to the Google Toolbar and to the static servers that it can be retrieved from.  Google search doesn’t use the static one. Duh.  Google gets free reign to use the dynamic PageRank.  Why? Because they make it.

They also get to change the algorithm at the whim of their engineers.  For that reason, the length of time since last update, and a few other reasons, there are many people who are calling for PageRank to be abandoned.  There are just as many that think it already has, we just haven’t been told about it.  I sincerely hope that both groups are wrong.

Why?  Because PageRank has become a standard.  There are many, many websites that use it as a metric for position, price, and power.  I must admit that using one metric like that is a bit on the wrong side.  It’s akin to only having one advertising revenue source on a website.  All well and good until that source stops working.

But we still can’t get rid of PageRank.  We need it.  It’s too much of a standard to just drop like a lead balloon.  What we need is an independent body that will conglomerate that and other metrics and then convert them into one, unifying, metric.  For the same reason that Alexa is frowned upon as a traffic gauge, we cannot depend on one metric alone.

What would my ideal metric include?

  • Pagerank
  • Alexa Rank
  • Compete Rank
  • Subscription rank (not currently possible as it’s not available on all blogs)
  • ?  Other’s could be integrated as they are developed.

How would the metric be calculated?

PageRank, Alexa, and Compete would be the core as they are the leading metrics.  The problem here is scale.  PageRank is on a scale of 1 to 10.  Alexa is a numeric rank based on position.  As is Compete.  Each is skewed a little in it’s own way and we can’t really play favorites, so we have to find a way to equally include each.  Obviously, we can’t just do an average of any sort.  That just wouldn’t work.  We need something that would scale.  Here’s how I would do it.

Take PageRank and multiply by 10.  Now it’s on a scale of 10-100.  Take Alexa and divide by 10,000.  Now it is roughly on a scale of .0001 to 300.   Compete is a little bit different.  There are several metrics that we can choose from, but the one we’ll use for this demonstration is Rank.  It’s a measure of time spent on the domain vs the time spent on the internet.  It’s a combination of a unique monthly count and the time spent on a site.  An “Engagement” metric.  Now we take the rank and divide it by 10,000.  Again, this gets us a rank between .0001 and 100 (Compete rank only uses the top 1 million sites based on unique people count).

Now we have a PageRank on a scale of 10-100, an Alexa score of .0001 to 300 and a Compete score from .0001 and 100.  Now, here we have a choice.  We can divide our answers and get a scale where the higher your score, the better, or we can multiply and the lower your score is the better.  We’re all about feeding egos around here, so I’m going to pick the divide answer.  Divide the PageRank score by the Alexa score and then divide the PageRank score by the Compete score and add the two together.  If you are unranked in both Alexa and Compete you end up with a score of 0 no matter what your PageRank is.  If you have a PageRank of 10 and are number 1 on both Alexa and Compete, you end up with a score of 2,000,000.  You might be thinking that that is a pretty large scale, but the beauty of it is that there is room for a lot of movement and a lot more detail in the ranking.

There are several other ways you could go about using those ranking metrics to come up with a unified ranking system.  Each with it’s own merits and with it’s own flaws.  What is important in creating a unified ranking system is that you attempt to keep things equal.  No one component should create a drastic skew in one direction or the other.  With this system I just explained, I think that is true.  (note: I may be wrong, as I wasn’t very good at math in school.)  But if your blog has a page rank of 10 somehow, but only shows an Alexa score of 1,350,789 and a Compete score of 999,145, you would only have a rank of  about 1.75.  As a comparison, this site has a PageRank of 5, a Compete score of 138,160 and an Alexa score of 159,484.  That give it a rank of about 0.68.  This site would still rank higher even though it only has half the PageRank.

I’m sure that someone, somewhere would find a way to game this or any other ranking system, but the more sources we have to rank with, the harder it becomes to skew the rank without gaming all of the sources.

What are your thoughts on ranking and metrics?

Technorati Tags: , , , , , , ,

About Shane Ede

Shane Ede is an IT guy by day and a Entrepreneurial Blogger by night. You can follow him here on Thatedeguy or over on Twitter and Google+.

Comments

  1. The main problem with Compete and Alexa is that the data itself is skewed: http://www.dgfmedia.net/blog/2007/05/15/alexa-traffic-ranking-idiosyncrasy/ (pardon the link drop). So, using it to form some other metric would be skewed by default (even though you’re adding PR in the mix).

    I agree that PR is something that should be kept around as a metric, but it keeps being degraded in favor of relevance (which is good). The value of PR is therefore a lot less now than it was say 5 years ago. Just wait and you will see that it will become about as useless as the keyword meta-tag is now.

  2. I won’t argue that they are skewed, and that they cannot be used as a metric on their own. However, due to the lack of anything better, we make some sacrifices and try and make it as un-skewed as possible by integrating several metrics to try and balance the load.

  3. The downside with Compete is that it only seems to use US visitors so anyone with a country specific domain name won’t necessarily get much ‘help’ from Compete. A few sites I tried are not coming up with accurate stats as they are UK domains.

    I agree, you need a good balance of metrics to be able to get a more accurate idea about a site. PageRank isn’t give to new domains until the update, Technorati authority doesn’t say anything about site visitors, simply how many people link to your posts, even RSS readers for a blog aren’t accurate from feedburner.

    I don’t understand why some of these advertising sites such as Text Link Ads can’t give you a piece of code that they could use to measure your site visitors and continue monitoring it. That would be the most accurate metric in my opinion.

Speak Your Mind

*