Monthly Archives: May 2009

How Does The Twitter Grader Ranking Work?

The most common question I get at grader.com is about how the Twitter Grader algorithm (and associated rankings) works.  Before we dig a bit into the details, it will help to understand the what before the how.  What Twitter Grader is trying to measure is the power, reach and authority of a twitter account.  In other words, when you tweet, what kind of an impact does it have?

Normally, we don’t like talking about the details of the Twitter Grader algorithm.  This is for the same reason that Google doesn’t like to talk about its algorithm: revealing details increases the degree to which people try to game the system.  So, lets approach the question from a different way.  If one were to look at data for a given user available in twitter, what kinds of things would one look at to determine whether that user had power, reach and authority?  Also, when looking at these various factors, it’s helpful to think about each of these in the “all other things being equal, what’s better” context.  Otherwise, it’s easy to get caught up into non-productive arguments on why a certain factor is or isn’t important, because there are so many cases that “prove” that it doesn’t matter.  Let me explain.  I’m going to tell you that one of the factors that goes into measuring your Twitter Grade is the number of followers you have.  Many of you would argue that the number of followers is completely irrelevant because it’s so easy to game that particular number using automated tools to do nothing but acquire followers.  I would counter with this:  If we were looking at two different twitter users, all other things being equal (and I do mean all other things), the one with more followers is likely more powerful and deserves a higher twitter grade.  Of course, all other things are usually not equal and that’s why the Twitter Grade is interesting.

So, let’s go into the factors.  Note:  These are NOT in order of priority or weight (and they’re not all weighted equally — not by a long shot).

Algorithm Factors

1.  Number of Followers:  More followers leads to a higher Twitter Grade (all other things being equal).  Yes, I agree that it’s easy to game this number, but we are looking at measuring reach and I did say all things being equal.

2.  Power of Followers:  If you have people with a high Twitter Grade following you, it counts more than those with a low Twitter Grade following you.  It’s a bit recursive, and we don’t get carried away with it, but it helps.

2.  Updates:  More updates generally leads to a higher grade — within reason.  This does not mean you should be tweeting like a manic squirrel cranked up on caffeine and sugar.  It won’t help either your Twitter Grade or your overall happiness in life.

3.  Update Recency:  Users that are more current (i.e. time elapsed since last tweet is low) generally get higher grades.

4.  Follower/Following Ratio:  The higher the ratio, the better.  However, the weight of this particular factor decreases as the user accrues points for other factors (so, once a user gets to a high level of followers or a high level of engagement, the Follower/Following ratio counts less).

5.  Engagement:  The more a given user’s tweets are being retweeted, the more times the user is being referenced or cited, the higher the twitter grade.  Further, the value of the engagement is higher based on who is being engaged.  If a user with a very high Twitter Grade retweets, it counts more than if a spammy account with a very low grade retweets.

The Grade Calculation:  So, those are the factors that go into the calculation of a score.  This score is then used to compare a user against all other users that also have a score.  The grade is calculated as the approximate percentage of other users that have an equal or lower score.  So, a Twitter Grade of 80 means that about 80% of the other users got a lower score.

The Ranking: The absolute ranking is exactly what it sounds like.  Based on all other users scored, what’s your “position” in that list.  A ranking of 5,000 means that only 4,999 other people had a higher score than you (at that point in time). 

Elite List:  The elite list is simply an ordered list of the top users (based on ranking) at a given point in time.  This list is updated several times a day.  We also maintain lists of the top ranking users based on a narrower set of users (like those in a specific geography, those that match a specific keyword, etc.). 

That’s all I’ve got for now.  Hopefully, this answers some of your questions.  What are other factors you think we should be looking at to compute the Twitter Grade?  Would love to hear your thoughts and ideas in the comments.