I Knew It Would Happen: Now We Can Really Measure Twitter

I’m doing a lot of thinking these days about measurement of the effectiveness of public relations programs.  We’re covering this in my class and my day job is, well, getting kicked around a bit of late.

I have long been a proponent of the premise that, in order to do good measurement, you need a “mashup” of tools.  You need to look at, of course, print, blogs, Web sites, message boards (especially in the world if finance), but measurement often lags behind the subject matter that it measures.

I’m coming late to the party, but ReadWriteWeb reported on the Twittermeter, a way to measure mentions in Twitter.  They state:

Enter Twittermeter. Twittermeter uses the Twitter API to scrape the site’s public feed and creates a database of every word sent over Twitter. Though database overages have forced the site to display only results for the past week, they have data since November 6th, 2007 totaling over 14.5 million words from 2.1 million status messages.

Twittermeter creates buzz graphs comparing words. For example, the graph below for the word “earthquake,” clearly shows a spike during the UK quake that took place earlier this week.”

Cool.  The challenge, for communicators, is now to add that to one big tent.  I am an unabashed fan of Custom Scoop, a platform that, while collecting information for thousands of print sites and blogs, also offers one of the opportunity to accept .xml feeds from other sites.  The more that you can measure under one big tent, the better. Tweetscan (or Twitter Search, whichever you call it) can also do it.

And while I am at it, measurement should not be about the tone or favorability ot articles, but of mentions of the company or issue that you are tracking.  Thanks to Katie Payne, I am now a disciple of “Measuring Public Relationships.




  1. Lindsey Brothers Says: October 15, 2008 at 4:08 pm

    But Mark, what about when Katie Payne writes, “The reality is that you need formal systems to track and evaluate (I’d italicize evaluate if I could) your entire program. Without them you will only be judged on your last headline.”

    Like you wrote in your blog, “measurement should not be about the tone or favorability of articles, but of mentions of the company or issue that you are tracking,” doesn’t this take us back to pre-modern-day-measurement? If we stuck with this method of measuring we wouldn’t know if 1,000 mentions of Starbucks were good or bad because we wouldn’t know how the tone or favorability measured up.

    I’m not a Twitter user, but I checked out Tweet Scan to see what some common discussions included. A term that many users had tweeted about was “Bank of America.” After scanning the multiple Twitter messages I found one saying, “Bank of America suck ass!” while another was, “Don’t know who your CC is through, but Bank of America made it very painless on me when my CC# was stolen somehow.” Wouldn’t you think that it is important to take into consideration tone and/or favorability?

    Going back to the “thump test,” you mentioned that organizations are now wanting to know more, not just how many hits they received. In other words, they want to know how favorable the article was to their brand.

    On a different note, I wasn’t aware of Twittermeter, I’m interested in seeing how this measurement tool develops over time. Thank you!

  2. Mark Story Says: October 15, 2008 at 4:13 pm

    Very good point about favorablity, but like many clipping services used to be, these tools, like Twitter, are in their infancy.


Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Follow by Email
%d bloggers like this: