January 03, 2011

Hashtag Analytics: A Twitter "Storm"

The weather folks in Portland, OR use #pdxtst to share information about weather-related events.  So whenever a "storm" is coming, you'll observe a "Twitter Storm"!

In late November, there was talk of a big storm coming to the Pacific Northwest.  In the four weeks prior to talk of a big storm, there wasn't much activity in the community:


Recency Cases Engage # Engage
1 Week 26 34.4% 9
2 Weeks 20 20.1% 4
3 Weeks 15 13.3% 2
4 Weeks 8 29.9% 2
2 Months 38 10.8% 4
3 Months 47 4.2% 2
4-6 Mo. 254 2.5% 6
7-9 Mo. 108 1.2% 1
10-12 Mo. 20 1.4% 0
Newbies 8 100.0% 8
Percent Newbies = 19.4%

Basically, there isn't much to talk about, and this is a closed community, without many new folks participating.

And then, the weather changes.  Take a look at how the metric change as a result!


Recency Cases Engage # Engage
1 Week 45 71.1% 32
2 Weeks 5 80.0% 4
3 Weeks 15 53.3% 8
4 Weeks 40 52.5% 21
2 Months 18 38.9% 7
3 Months 39 33.3% 13
4-6 Mo. 221 19.0% 42
7-9 Mo. 140 20.7% 29
10-12 Mo. 27 7.4% 2
Newbies 292 100.0% 292
Percent Newbies = 64.9%

Pow!!

Notice that engagement rates go bonkers, regardless of recency since last tweet.  The entire population is buzzing about this event!  Even more important, nearly two out of every three "tweeters" are new to the #pdxtst.


Now, we need to see what happens the following week.  Take a look at engagement rates.



Recency Cases Engage # Engage
1 Week 450 6.9% 31
2 Weeks 13 0.0% 0
3 Weeks 1 0.0% 0
4 Weeks 7 0.0% 0
2 Months 21 0.0% 0
3 Months 32 0.0% 0
4-6 Mo. 137 0.0% 0
7-9 Mo. 150 0.7% 1
10-12 Mo. 31 0.0% 0
Newbies 5 100.0% 5
Percent Newbies = 13.5%


Clearly, there wasn't anything worth talking about, so the community shut down.


The following week, there's a bit more buzz, but only among recent participants.



Recency Cases Engage # Engage
1 Week 37 35.1% 13
2 Weeks 419 2.6% 11
3 Weeks 13 0.0% 0
4 Weeks 1 0.0% 0
2 Months 28 0.0% 0
3 Months 28 7.1% 2
4-6 Mo. 133 0.8% 1
7-9 Mo. 154 0.0% 0
10-12 Mo. 34 0.0% 0
Newbies 8 100.0% 8
Percent Newbies = 22.8%


I'm going to show you what happens for one more week.  Pay attention to the glut of folks who participated during the winter weather event ... they've dropped down to three weeks of recency.



Recency Cases Engage # Engage
1 Week 35 54.3% 19
2 Weeks 24 37.5% 9
3 Weeks 408 15.9% 65
4 Weeks 13 15.4% 2
2 Months 27 11.1% 3
3 Months 21 4.8% 1
4-6 Mo. 107 7.5% 8
7-9 Mo. 175 1.7% 3
10-12 Mo. 45 4.4% 2
Newbies 58 100.0% 58
Percent Newbies = 34.1%



Ok, there's weather to talk about this week, and the audience comes back --- but most important, there's the glut of people with three weeks of recency ... those folks engage at a normal rate, but there are so many people in this cohort that they fuel the conversation ... sixty-five people in this band engage about lousy weather.


What's the point of this analysis?


There are two key takeaways:
  1. When something goes viral, you end up with a large cohort of individuals who have the potential to fuel a conversation in the future.
  2. For there to be a conversation, there has to be something to talk about.  Notice how this community comes to life when there is something to talk about!
In other words, a significant weather event "awakened" this community, fueling high engagement rates, and fueling a ginormous increase in new users.  But a ton of new users/followers/fans doesn't mean anything unless this audience has something to talk about.  Counts and numbers have very little meaning ... having something to talk about is important.


Want to learn more about Hashtag Analytics?  Give Hillstrom's Hashtag Analytics a read!