My friend and independent musician Drew LaPlante likes to use the phrase, “It’s a beautiful thing.” I could say that about Twitter and Twitter search.
Using Twitter Search to track trends
One way I use Twitter is to track the presence of certain phrases over time. I subscribe to the RSS feeds for searches in Google Reader and let them sit for at least 30 days. I’m not as much interested in the content of the individual Twitter updates(tweets) as the aggregated information. I find this exercise to be instructive in understanding trends.
What’s being said and how I might help
I had thought of doing this to identify opportunities to promote the How I Got My Job job-hunting success stories podcast. My original hypothesis was that people might be more interested in knowing about the series right when they’ve lost their jobs.
Understanding the stories data can tell
In all the data analysis I’ve ever done at work and in other projects, to better understand the stories data tell, I’ve realized it’s vital to check and verify the underlying data, and make sure assumptions are clear. That way I and others can make better business decisions.
Analyzing Twitter search trends definitely comes with a caveat. Take this snapshot of Twitter search data, as shown by day of the week, for example:
This chart shows that, over the past 30 days, the phrases “just lost my job” and “lost my job today” have showed up on the Twitter public timeline on Wednesday the most. Does this mean that more people are losing their jobs on Wednesdays the most? Maybe, but you cannot be sure from this visual alone.
Why? Because aggregate information like this, a gross overview of the presence of phrases or terms, ignores context. What does “just lost my job” mean to you? Does it mean today? Last week? A couple weeks ago? What about those who do not report their job loss on Twitter with these words?
Even so, Twitter search trend analysis like this is instructive to know that people are talking about losing jobs, and it identifies opportunities to share about the series.
The visual above shows the same Twitter search data over the past 30 days, but in this case by time of day. The orange bar shows what time I’ve marked the items as read in Google Reader and can be ignored. The chart indicates more people mentioning “lost my job today” or “just lost my job” in the afternoon or evening, perhaps, after coming home from their last day of work.
How I use the data
So, what does this mean for me? It seems that promoting the show on Twitter in the afternoon would perhaps reach the largest audience that might be most receptive, and maybe doing this on Wednesday would be best. I still need to find a way to measure whether these efforts are successful, and this could be done by checking web analytics.
What say you?
I don’t claim that this approach is without flaws, and would love to see how it could be improved. I would like to better track the effectiveness of promotions and welcome your thoughts. What do you think of this approach? What would you do differently? Have you ever taken a similar approach?
April 13, 2001 UPDATE: Christopher Penn shares some additional insight into timing Twitter updates that you should read. Check out “When is the best time to tweet?”
If you liked this post, click here to share it with your network!
Want to read more like this from Daniel Johnson, Jr.? If you do, subscribe right now!
Get via Email 
Add to Google 
Subscribe to the Feed
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_e.png?x-id=6b1ef753-e162-41ca-af2d-bb2ed4b6dcd0)
![Reblog this post [with Zemanta]](http://img.zemanta.com/reblog_e.png?x-id=aaec6bf5-8425-41ad-bd4b-2c639cb78d21)
