Alumni Associations Twitter Usage Report

Don't let the fact that I am not able to offer you a fully functioning version of this data with an attached, preloaded pivot table stop you. I would love to put a call out there for someone to do their magic with this data and give it back to us. I ventured into this project with a few basic questions I wanted to answer. I collected the data and set about to answer those questions. There is, without a doubt, a hundred other questions and answers that one could derive from this data, so have at it.

If you plan to do your own analysis and discover some answers to your own questions, you should know a few basics about the tool I used to collect this data. Twitalyzer is a website that runs the numbers you see represented in the data. Twitalyzer takes a snapshot of the past seven days of your use of twitter and runs the algorithm to arrive at a score in 5 different categories, influence, signal, generosity, velocity and clout.

A couple other things you should know...If you haven’t used twitter in the past seven days, it doesn’t matter how much influence or clout you had, Twitalyzer will give a zero return on data. This does NOT mean that your influence scores will reset, start over from zero, and need to build up again. Part of the Twitalyzer algorithm accounts for the number of followers you have, and previous profile scores that have been analyzed in the past. It is important to understand how some profiles might go from an influence score of 2.4 to zero and back to 2.2. The “no data” score is not technically a score of zero, even though it might seem that way in how you choose to present data. Twitalyzer basically answers the question, “what is your influence in twitter in the last seven days.” Where one can easily make the case that a profile that hasn’t been updated for the past week still retains some influence (based upon user accessibility to previous updates), Twitalyzer only shows trends limited to the previous tabulations that have been run on a given user name. In the basic report page, an ongoing tabulation of the scores is only presented in terms of averages… and not trends. If you would like to see the information presented in a line graph, over time, Twitalyzer will give you a look at that information in a couple of fun ways here.

Based upon this information, it seems inappropriate to present these numbers in a line graph, over time.

This raw Twitalyzer data should be seen as a snapshot average, tabulated five times, between the dates of May 27th and June 22nd , 2009.

If any one data collection time returned “no data” that time was not counted as a zero in the average, instead it was not included in the average.

So, lets get into the questions and answers about Alumni Associations.

I took time to select the top 40 twitter profiles (based on number of followers) by searching for user bios or user names that contained the words "alumni association." the remaining group of alumni association feeds came from a sampling of feeds that were interesting to me for one reason or another. For instance, the two profiles at the bottom of the list were selected because they were started within a couple of days of when my data collection began.

What does the average alumni association feed look like?
After collecting data over the course of a month, what was the aggregate growth seen in each time period?
The typical twitter profile for alumni associations started off with 454 followers and following 303 twitter users. Those numbers grew. By the end of the month, the typical alumni association profile was following 356 users while being followed by 559.

Lets average all the scores together for all 49 profiles that were traced over the course of the month.

The fact that alumni associations have an average influence score 0.4 give us plenty of comparisons to make among the ones that were analyzed. Should alumni associations be happy to be above average? Do their twitter feeds rank low in comparison to other similar industries? More on that later.
For now, here is several quick looks at how individual profiles compare.




















If you have been reading these posts from the beginning, you have heard me make the case that judging a profile based on the number of followers is a horrible way to measure success. It is still a statistic that most would be interested in. Here are a couple of ways to look at that statistic. There are a couple of ways to determine how engaging an alumni association is being with this social web tool. One quick way I did this was to look back through three pages of updates for each feed and see if they had a tendency to @reply any of their followers with conversation. If there were none, it is easy to conclude that a feed was dedicated to one-way communication. Those profiles that choose to engage with @replies saw slightly higher growth in followers than those who did not. While I personally hold to the insight that it is better to have 50 responsive followers than 1000 that ignore you, this data comparison can help one make several observations.

There also seems to be a relationship between high clout/influence scores with those profiles who make a practice of following a high percentage of users that are following them. This is seen, in twitter, as a more generous and engaging way to act as a user.

The question becomes, to the alumni association, if the simple act of following back the real people who are following your feed is worth the little bit of extra time it takes to do this. The influence scores say yes.

You can see that the majority of alumni associations fail to take time to enact a follow strategy that leads out and initiates a relationship with another user by following first. Those that do, and follow at least 90% of their following, have more than three times the influence and clout of those that follow fewer than 50%.

While I have several other observations to make about this data. This might be enough to get our conversation started for now. What do you think? Feel free to pull down the spreadsheet and run your own numbers? What does it tell us? What recommendations can we make?
More to come...