Larry Brauner
Everything defies measurement and tracking. Everything, really.

Try recording your food intake. If you’re an emotional eater, committing your diet to paper ought to make you feel quite uncomfortable. However, even if you eat normally, listing your meals poses the following interesting problem:

Because you’re making a list of your meals and snacks, you’ll tend to make healthier eating choices than you would otherwise make. You may eat less or more than usual. Therefore, the items on your list will not represent your typical food intake. Rather, they’ll be biased.

Tracking the Untrackable

For several years, I worked as a business analyst at IDT Corporation.

One of several areas responsibility at IDT was analyzing advertising tracking data, partly in order to evaluate each advertising purchase, and partly to in order to determine the net value of each customer (after factoring out the cost of acquiring that customer).

Given our level of sophistication, tracking the new customers by marketing channel should have been straightforward. Nevertheless, there was a major problem: Our new customer defied tracking.

A different toll-free telephone number and a different web address was used for each newspaper, magazine, television station, radio station and direct mail piece. When a customer called the toll-free number or visited the web site, we knew how the customer was referred to us.

Well, sort of. The problem was that the customer didn’t always behave as we had hoped.

No matter which phone number or web address the customer was given, that customer sometimes found it more convenient to obtain the phone number by calling Information or going online, and to bypass their assigned web address, going to the company’s main website instead.

We called such a customer “untrackable” and were forced to make the best assumptions we could to deal with the untrackables in our analyses.

Tracking Twitter

I can provide many examples of tracking and measurement difficulties, especially from my years working in marketing research at Eric Marder Associates, but not to bore you too much, I’ll jump now to my discussion of Twitter.

My thoughts on Twitter will apply in varying degrees to Facebook and other social sites as well. I break down the measurement and tracking of Twitter traffic into these eight parts:

  1. You need to realize that much activity on the Internet, and on Twitter specifically is generated by cyber robots or plain bots. They tweet the majority of updates on Twitter, and they account for more than 90% of the traffic that flows through the links in Twitter posts.
  2. While some techies may be very interested in bot activity, most of us are simply interested in counting and tracking human clicks on our links. We need to separate out and count only real clicks by real peeps.
  3. Realize too that most humans access Twitter from desktop and mobile clients, not from the Twitter domain. (The extent to which this is true depends on the particular audience you’re targeting.)
  4. Web analytics tools, such as Google Analytics and Clicky, do exclude bot traffic from their stats. However, they do not know how to break down and allocate the so-called direct traffic coming from users’ desktops and mobile devices. In web analytics, direct traffic is the untrackable element which I discussed above in connection with my work at IDT. Nevertheless, do not rely on the stats from your site’s log. Install and use Google Analytics and Clicky in your blog or website. I use both myself. (For my Wordpress blog, I use the Ultimate Google Analytics plugin, and I installed the Clicky script in a sidebar.)
  5. Web untrackables come to websites in many different ways, such as directly typing a website address, selecting browser bookmarks, using a variety of desktop and mobile clients like TweetDeck, and even clicking on a link in an e-mail program such as Microsoft Outlook. Try and get a sense of where your direct traffic comes from.
  6. Rules of thumb provide no more than ball park estimates, and these crude approximations are often inadequate. Use rules of thumb only as a last resort.
  7. You can substitute tracked links in your tweets, but tracked links generally count bot traffic. However, BudUrl from Live Oak 360 has begun counting only human clicks. Great news! While their links can’t be generated automatically by TweetDeck, if you’re serious about tracking, you’ll put up with the inconvenience. Not every tweet will have a link, and not every link will need to be tracked.
  8. Even if you use BudUrl as I recommend, there’s still one more thing defying measurement, the Twitter user who replaces your link with theirs in order to track their retweet or because they prefer another shortening link. No way around this one! Remember, “Everything defies measurement and tracking.”

I count the comments made on each article. Don’t you defy measurement and tracking. :-P

Share your ideas below in a comment. :-)

Did you enjoy this article? Subscribe to my RSS feed or by e-mail. Visit my About, Services, Media Buzz and Connect pages to learn about Building Your Audience and Brand on the Web. See also my Disclosure Policy regarding affiliations and compensation.

You Comment I Follow Online Social Networking Home Online Social Networking Sitemap About Larry Brauner
Sharing is Caring!
Tags: , ,


If you found this page useful, consider linking to it.
Simply copy and paste the code below into your web site (Ctrl+C to copy)
It will look like this: Twitter Stats Defy Measurement

7 Responses to “Twitter Stats Defy Measurement”

  1. Debbie Morgan on November 25th, 2009 6:56 am

    Hi Larry,

    I thought I was the only one that couldn’t nail this concept down. I always have a large number of visitors in the “unknown” category. I’m glad they found me, but it would be nice to know how they found me. Thanks, Larry, for a great post.


  2. Chris Boyer on November 25th, 2009 9:21 am

    Larry, another good blog post. The topic of measurement of ROI in social media is always popping up, and blog posts like this provide real tactics on how to get started.

    I remember way back before Google came on the scene, people were struggling with measuring website traffic. Now we have great tools (Google Analytics, Webmaster Tools, Clicky)…I am sure we’ll start to see a more rigorous suite of tools develop for SM. Until then, posts like yours are very helpful. Thanks!

  3. Larry Brauner on November 25th, 2009 10:21 am

    Thank you Debbie and Chris. I look forward to new too. I’m sure I’ll still find enough problems to solve and topics to write about. ;-)

  4. Kimberly on November 25th, 2009 4:11 pm

    Good post. Helpful information here. I try to do some pretty rudimentary measurement on a monthly basis for my clients. I do like BUD url b/c you can exclude the bot counts but I find their interface cumbersome and prefer some of the features of for link tracking. I also sometimes use and for just measuring some more of the basic for overall growth and tweet habits. Not ROI for sure but I think being able to have some data is helpful for those new to using social media. I also like to take Facebook Insights data and match it with the items posted throughout the month to get an idea what types of posted items where more *sticky* than others. You might also want to sign up for a beta invite to as they are developing some pretty cool Twitter measurement tools.

  5. Bruno on November 27th, 2009 10:12 am

    Hi Larry,

    I have the same problem with my lead generating websites. I have a network of moving sites and the phone number versus leads creates a real problem.

    So what I do is rent the phone number to moving companies per month, not very trackable but kind of works. :(

  6. Larry Brauner on November 27th, 2009 10:51 am

    Everybody Bruno, who has made a serious attempt to track their leads will tell you that their tracking system has leaks. All we can really do is to minimize and mitigate those leaks.

    Thank you for your comment.

  7. My Top 10+ Blog Traffic Sources on January 25th, 2010 10:40 am

    […] – I’ve written at length about Twitter. Read Twitter Stats Defy Measurement. I’m happy to have started with Twitter in 2008 when Twitter’s rules didn’t get in the way of […]

Leave a Reply