If you click an outbound link on Twitter.com and immediately hit stop, your browser’s address bar will display something something similar to:

http://twitter.com/link_click_count?url=http%3A%2F%2Fsu.pr%2F1VF3eU&linkType=web&tweetId=3822962256&userId=17594917&authenticity_token=fa92ae1a3b002f6051a9f8d304cef09e0095ee09

If you allow this link to continue loading, you will be redirected to the URL on which you clicked via a 0.1 second meta refresh.

Let’s take a look at what information Twitter is collecting.

http://twitter.com/link_click_count

It appears fairly obvious that Twitter’s aim is to count clicks on outbound links.

url=http%3A%2F%2Fsu.pr%2F1VF3eU

If you’re not familiar with URL encoding, don’t let the messy-looking characters scare you. This is the equivalent of http://su.pr/1VF3eU

linkType=web

Refers to where the link was clicked. At this time, it appears that Twitter is only tracking links on Twitter.com. The existence of this parameter seems to be evidence that they are considering expanding beyond that.

tweetId=3822962256

A unique number assigned to every Tweet. Combining this number with the username of its Tweeter will display the status update: http://twitter.com/seanwf/status/3822962256

userId=17594917

This number is used to uniquely identify each Twitter user; roughly equivalent to their username.

authenticity_token=fa92ae1a3b002f6051a9f8d304cef09e0095ee09

A unique random string value. This security measure was unfortunately exploited by the hacker behind the StalkDaily Worm fiasco.

What Does Twitter Know?

We can glean from this that Twitter now has access to a smorgasbord of information about links on Twitter.com. As it turns out, the “authenticity_token” mentioned above is actually a cookie-based string. What this means to us is that Twitter can track a user’s click-behavior over multiple accounts.

Twitter can correlate your click-behavior with all of the information related to your account including:

• Name
• Location
• Bio
• Homepage
• Following
• Followers
• Date Joined
• Time Zone
• Tweets

The same data is available about the user who shared the link with you and any users mentioned in the tweet. All data related to the Tweet containing the link is also tied to your click such as:

• Date Created
• Users Mentioned
• Application Used

Good for Twitter, Bad for Users

Personally, I have noticed that this redirection adds a noticeable lag when loading external pages. It appears that I am not the only one experiencing this unfortunate side-effect. Another negative consequence to users is that when Twitter experiences server issues, the outbound links on Twitter.com may stop working.

Twitter has yet to announce any plans to make this data available to developers. Could this be part of a larger strategy to make Twitter.com relevant despite less than 10% of the platform’s traffic passing through the website? One thing is certain. This new feature gives Twitter access to an enormous quantity of valuable information that can be used to generate revenue through their new commercial accounts.

Delving Deeper into the Data – Twitter Demographics

In my next post, we will be exploring the demographics of Twitter users through a meta-analysis of about a dozen data sets. What we learn may surprise you and challenge some assumptions about the types of people using the micro-blogging platform.

Notes

Photo curteosy of Flickr and:

Related Posts:

• Social Media for Small Business – Yoono Models Best Practices
• Researching the Validity of Online Competitive Intelligence Tools – Introduction
• 10 Reasons You MUST Attend IZEAfest 2009

Twitter’s “Link Click Count” Provides Analytics, Revenue, & 503 Errors

was brought to you by WellonTop Internet Research

StumbleUpon v. Digg – A Case Study

The purpose of this case study is to describe the predictions made by five free competitive analysis tools. We will do this by comparing estimates of unique visitors to two popular social bookmarking websites. At the end I’ll share a small but valuable secret that will allow us to test any hypotheses we form.

Quantcast

A cursory glance at Quantcast’s visitor estimates for StumbleUpon.com isn’t particularly interesting. According to Quantcast, StumbleUpon receives less than a million unique US visitors/month. Even worse, the website has shown a dramatic decline in traffic since March of this year. For a website with nearly 8 million members worldwide, StumbleUpon’s numbers are at best disappointing, especially compared to Digg’s 12.5 million unique US visitors/month.

Compete.com

Comparing StumbleUpon and Digg on Compete.com tells a slightly different story. In strong contrast to Quantcast’s directly measured data, Compete estimates Digg receives almost 39 million unique US visitors/month. Compete’s estimate is more than 3x the actual traffic volume. Compete also comparitively exaggerates StumbleUpon’s traffic at 5 million visitors; more than 6x Quantcast’s estimate.

Alexa

Adding another tool only makes things more confusing. Alexa estimates the global reach of Digg and StumbleUpon to be .53% and .19% respectively. It’s important to note that we cannot directly compare Alexa’s “daily reach” (even averaged over 30 days) with “unique monthly visitors.” The problem is that each measure two distinct (however related) variables. In order to properly compare the two, we need to account for how many visits unique visitors are responsible for. If that doesn’t make sense yet, don’t worry, I still have a hard time wrapping my head around it too. Leave a comment and I’ll make sure to follow up with a post explaining it.

For the purposes of this limited case study we’ll assume there are approximately 1.6 billion internet users worldwide, half of whom (800 million) are online on any given day. Digg should be receiving 4.2 million unique global visitors/day compared to StumbleUpon’s 1.5 million. We know that Alexa’s estimate for Digg is more than double the true number, but what about StumbleUpon?

Google Trends

At this point, it seems like we may as well just throw our hands up in the air and acknowledge that free competitive analysis estimates are largely worthless. Fortunately, at WoT, the words “giving” and “up” don’t hang out together. They’re much more fun by themselves anyway. Luckily for us we stumbled upon (punny) something spectacular in the process.

As usual, data-superhero Google comes flying to our rescue. Comparing Digg and StumbleUpon on Google Trends begins to clear up this mess. At the end of March, StumbleUpon has less than half the unique US visitors/month of Digg. The gap closes quite suddenly in mid-April in part due via a combination of three major events:

1. April 13^th – StumbleUpon is bought back from eBay by its founders
2. April 22^nd – Version 3.29 of the StumbleUpon Firefox toolbar is released
3. April 27^th – Web Stumbling (without the toolbar) receives major enhancements

Competitive Analysis Tools Compared

Not all of our competitive intelligence tools have data past May. To compensate we will only use the month of April, identified by the period immediately prior to StumbleUpon’s traffic spike.

Immediately we see the degree to which Compete has egregiously overestimated Digg’s traffic. Google Trends is almost certainly an underestimation. What about the data on global visitors?

Alexa and Google Trends both drastically underestimate Digg’s global traffic. While we don’t have any way to know for sure, I would posit that they also underestimate StumbleUpon’s data.

Interpreting Results

So what exactly can we take away from this case study? For one, we can be pretty confident that StumbleUpon received significantly fewer unique global visitors in the first few weeks of April than Digg. We can even take a stab at the difference. Assuming that StumbleUpon data is skewed in proportion to Digg data biases, our corrected global estimate for Google Trends is 8.4 million users. Applying the same methodology to Alexa gives a global visitor estimate of 7.3 million.

We know that about 50% of Digg’s unique users are from outside the US. Google Trends’ estimates can account for 84% of global traffic but only 53% of US traffic. To me this indicates a slight measurement bias in favor of international users. If StumbleUpon’s Google Trends data is similarly biased, our corrected US traffic estimate is 4.4 million users. When we apply the same technique Compete, the estimate falls from 5.9 million to 3.9 million users. Quantcast’s corrected estimate seems somewhat low at 2.2 million users.

Forming Hypotheses

It’s fairly surprising just how consistent our corrected data is. Based on the information gathered, I’m going to hypothesize that during the measured time period:

• StumbleUpon received between 7 to 9 million unique global visitors
• StumbleUpon received between 2 to 5 million unique US visitors
• Free competitive analysis tools can be useful if biases can be compensated for

A Special Bonus

In this case, I can test the first two hypotheses. I found a very valuable little snippet at the bottom of StumbleUpon’s source code. I’ll give you a hint: It begins with <!–Start Quantcast tag–>

Secret StumbleUpon Data on Quantcast!

Notes:

Some charts on this page have been modified. Links to the original sources have been included.

Photo: kevinzhengli on flickr.com

Related Posts:

• Website Traffic – Are Competitive Intelligence Tools Accurate? (Survey)
• A New Study on Stephen R. Covey’s 7 Habits of Highly Effective People
• Researching the Validity of Online Competitive Intelligence Tools – Introduction

Competitive Intelligence Tools – Alexa, Compete, Google Trends & Quantcast

was brought to you by WellonTop Internet Research