Getting a Handle on Twitter
Our last post examined the social media networks that higher education institutions use, the control exercised over accounts and links to networks from a website.
This post analyses and benchmarks six aspects of Twitter:
- Age profile of Twitter accounts;
- The number of tweets being produced;
- The tweet rate for Twitter accounts;
- Number of followers for an account;
- Number of accounts being followed; and,
- The follower:following ratio.
We selected Twitter for more detailed review as it is ubiquitous and offers an ease-to-use API to extract account details from its data store. Our analysis uses data for the ‘official’ Twitter accounts for 150 Canadian universities and colleges obtained from Twitter on April 20, 2016.
In this post, the graphs tell the story. Our commentary provides context.
Twitter Account Age Analysis
The first tweets and Twitter accounts date from March 2006 (click on the link and try @Jack). The oldest account in our sample group was opened in August 2007: while this institution was an early adopter it is no longer a very active account.
The most intense period for joining the Twitter bandwagon occurred between October 2008 and December 2009. During this period about 50% of all institutions opened their official institutional Twitter account.
Account opening was substantially complete by the end of 2013, with a few laggards joining Twitter in 2015 and 2016. Overall, 90% of the sample group of post-secondary institutions were active on Twitter by June 2012: most institutions have had at least four years to come to terms with Twitter and determine its effectiveness as a communications medium.
Graph 1: Twitter Account Opening Dates for Higher Education Twitter Accounts [proportion of sample group accounts opened in a calendar quarter]
Total Number of Tweets
Before we started crunching the numbers, our first presumption was that an account’s total number of tweets would be correlated to the length of time the account had been in existence. In fact, regression analysis shows that the age of an account only explains 10% of the change in the total number of tweets (for the statistically minded R2 = 0.103). In other words, if volume matters, getting a head start doesn’t make a difference. The largest numbers of tweets have been produced by the accounts that opened in the period October 2008 to December 2009.
We did not perform a statistical analysis of the relationship between the size of an institution (total student population) and the number of tweets, but scanning a list of the institutions ranked by number of tweets strongly suggests that the larger the institution the greater the output.
Graph 2: Tweet Output Histogram For Higher Education Institutions [proportion of accounts in each output band - i.e. 14.3% of accounts have generated between 3,500 and 4,500 tweets]
The data for the total number of tweets suggests three potential Twitter strategies. Some institutions have reserved their official Twitter account for low volume, but ‘strategic’ tweets. Another group uses the account more frequently and for a distinct set of topics. A third group tweets news, events and updates throughout the day. Combining the results of Twitter Analytics and Google Analytics would provide insights into which of these approaches is most effective in bringing relevant visitors to a website. Any feedback or comments from social media practitioners on this topic would be welcome.
We also reviewed the variation in the rate at which accounts tweets. To do this we took the total number of tweets produced and divided by the number of days the account had been in existence, as at 20 April 2016. The calculated rates show the range of tweet frequency and give a sense of how often a follower of a higher education Twitter account might expect to see new tweets in their timeline.
Graph 3: Twitter Rate Analysis for Higher Education Accounts - tweets/day [proportion of accounts in each rate band - i.e. 17.0% of accounts tweet between 1.5 and 2 times per day]
The ‘average’ higher education account tweets between 2 and 3 times a day (2.5 times to be precise). However, the data is skewed by a small number of active accounts. In practice, 35% of accounts tweet less than once daily.
While a third of the accounts can be considered low volume, the balance tweet at least four times a day. We assume that the rate is closer to one tweet every three hours, based on a 12-hour daily Twitter operating window.
Among our sample group we found a sub-set of five per cent of the accounts that tweet at least 10 times per day: which is likely closer to hourly for the ‘working day’.
There is significant dispersion in the total number of followers for the 150 accounts we reviewed: from fewer than 50 to a little over 76,000. Using an arithmetic average the ‘typical’ Canadian university or college account has 8,556 followers. The relative weighting of a small number of accounts with large numbers of followers, means that it is more relevant to note that sixty per cent of our sample group have fewer than 4,000 followers.
Graph 4: Twitter Followers Histogram For Higher Education Institutions [proportion of accounts in each band of follower counts - i.e. 57.8% of accounts have fewer than 4,000 followers]
Only five per cent of the accounts reviewed have more than 40,000 followers. These are the most active accounts in the review group and all opened in the October 2008 to December 2009 period.
Unfortunately, Twitter does not provide access to time and date data that permits analysis of the rate at which accounts attract followers. It may be reasonable to assume that, in the absence of active promotion, after a period of four years (recalling that 90% of the accounts studied became active four or more years ago) the follower base reaches a plateau and relatively few new followers are being attracted. Moreover, as relatively few individuals ‘clean up’ their social media presence, it is likely that many followers of higher education accounts no longer actively follow the official Twitter account after graduation: they may shift attention to an alumni account? It would be beneficial to segment the follower base to determine what portion of the audience is ‘actively’ addressable and what portion has simply not got around to unfollowing.
It is an active choice to follow a Twitter account. Individual institutions may argue with our conclusions, but there appears to be no clear strategy governing how many and which other accounts are followed by official university and college Twitter accounts. As a result, there is substantial variation in the number of accounts being followed and in the ratio of followers to following.
Graph 5: Number of Accounts Higher Education Twitter Accounts Follow [proportion of accounts in each band of following counts - i.e. 49.0% of accounts follow fewer than 600 accounts]
The ‘average’ Canadian university or college Twitter account follows 1,066 other accounts, but this overstates the case. Approximately half the sample group followed fewer than 600 other Twitter accounts. And, even the most active (by number of tweets) account in the group only followed a little over 15,800 other Twitter accounts.
As universities and colleges are highly decentralised entities it must be difficult to ensure that an official account identifies all the accounts belonging to departments, institutes, faculty, staff, students and alumni that could be relevant and might be appropriate to follow. As a result, for many institutions the choice of accounts being followed makes interesting reading.
Given the observations about relative numbers of followers and accounts being followed, we calculated the follower-to-following ratio for our sample group.
Generally speaking a high ratio of followers to following is deemed good, as it is strongly indicative that an account is producing tweets that attract an audience. However, a high follower-to-following ratio also suggests that Twitter is being used to broadcast rather than converse.
>Graph 6: Ratio of Followers to Following for Higher Education Twitter Accounts [proportion of accounts in each rate band - i.e. 61.6% of accounts have less than an 8 to 1 ratio of followers to following]
Outside the university and college world there are many active accounts where the ratio is close to one: these accounts operate a policy of 'following back'. We are not convinced of the merits of this approach, after all, one is judged by the company one keeps and it can be hard to identify the ‘quality’ of one’s company on Twitter.
The average post-secondary account has seventeen times as many followers as accounts it is following. But, we found that about sixty per cent of the sample group had a follower-to-following ratio of eight times or less: that is eight followers for every one account being followed. A very small sub-group actually follows more accounts than they have followers: a characteristic associated with nascent accounts still finding an audience. This set of accounts is associated with smaller educational institutions that, upon further examination, appear to have ‘lost interest’ in Twitter. At the extreme end of the scale, the ratio climbs to over 100 times for the top three per cent of accounts.
Twitter has been widely adopted by higher education institutions albeit with varying degrees of enthusiasm. The benchmarks that our review has identified are interesting as a guide to general practice, but the more interesting question to be answered is what outcomes are being sought by those institutions actively using Twitter?