Analysis of my 2016 Tweets

I have now been a Twitter user for 10 years per this tweet.

It's my #Twitterversary! I have been on Twitter for 10 years (since 12 Dec 2006).
And you? https://t.co/aMERj7Z4vC

— Simon Fogg (@srjf) December 12, 2016

Prior to this year, I have previously only textually summarised once my use of Twitter in this blog post.

2016 has been the most productive year for my learning from other people on Twitter. The main examples include participating in various TweetChats and tagging along virtually to a number of conferences. In both these cases, these are tribes that I do not necessarily feel at home in given my current professional role as a project manager, requirements/business consultant, service manager etc. I am learning to further expand my echo chamber and hopefully contributing myself too.

I am looking to have One Word as my theme for 2017 and it may be that the word will be “discipline” which will include me being far more intentional about what I am doing.

So as 2016 ended, I decided to analyse my tweets over the year.

One of the drivers was to see if I could confirm that I am a “giver” rather than a “taker” or a “matcher” per Adam Grant's classifications. I like to categorise myself as “an inveterate sharer of links”.

Compiling the Stats

I have no access to flashy tools and I am not personally aware of any tools that classify tweets in the way that I was hoping to.

Worth me saying that I had no fixed goal in mind so I basically made this up as I went along.

This is what I did:-

1. Obtained the latest version of my Twitter archive (instructions).
2. From the resulting zip file from Twitter, I used the csv file in Excel, deleted the tweets from before 2016 and added two columns:-
   1. Audience: with valid values (eventually!) of:-
      1. Named Recipient: for Tweets which had 1 or more named Twitter users referenced as a recipient
      2. No Named Recipients: for Tweets where there was no named recipient and were simply tweets about what I was up to, watching, doing etc
   2. Classification: valid values (eventually!) of:-
      1. General Chat: for Tweets that are social chat, asking questions, answering questions with no links to content included
      2. Sharing Links: for Tweets where I was sharing links to videos, books, music albums, TV programmes, web pages etc
      3. Links to Music/Podcast/Radio/TV/Video (usually that I was playing at the time)
      4. TweetChat #pkmchat
      5. TweetChat #ldinsight
      6. TweetChat #leadershour
      7. TweetChat #esnchat
3. I started going through January tweets and it took me 1 hour to classify them individually. I could not afford the time to do that amount of work for the remainder of the tweets.
4. So I started classifying tweets by audience, looking for tweets starting with “@” or “.@” which would be Named Recipients and the remainder as No Named Riecipients.
5. It is a bit hazy now in the cold light of day but I then went through iterations of identifying No Named Recipients tweets where I was simply saying what I was watching, playing etc, what were General Chat and then which were related to specific TweetChats and what were Sharing Links.
6. At the start of this exercise, I thought the split of no vs named recipient was the primary split that I was after but during the secondary classification analysis that became more important to me and I produced a pivot table in Excel (see graphic below).
7. At this point I realised that I had tweeted extensively during the #SocMedHE16 conference (virtually) and I added that info to the relevant labels (couldn’t face doing the split in Excel after the earlier work!).
8. I then took a screenshot and posted this tweet:

my Tweet stats/classifications for 2016; any feedback, questions, suggestions are actively encouraged & will be gratefully received #wol pic.twitter.com/JBLujxvB6e

— Simon Fogg (@srjf) 23 December 2016

Comments from doing this exercise

1. As per the screenshot, “Classifying 10.1k tweets manually was a challenge and consequently some simplifying assumptions were made to enable some automation! This was my best rapid shot.”. As an aside, I now do not have the detailed Excel classifications as it looks like I did not save it which at least means I cannot revisit it!
2. It made me think for the first time how I would want to classify tweets moving forward. I am confident that there are major simplifications in how I classified the Tweets. Presumably, there are lots of useful tweets in Social Chat where I am responding to people’s questions or asking questions of others rather than social chit chat as in pub talk and the like.
3. It also made me wonder how this could be made easier next time round. I have no immediate answers apart from splitting the work by doing it monthly rather than at the end of the year when I am in holiday mode and my brain is in rest mode as a result.
4. I was pleased that already 7% of my tweets are TweetChat related having only startd these recently.
5. Delighted that c55% of my tweets are sharing links to content so at least the bulk of my tweets are actually sharing links and hopefully helping others. So confirming that I am a “giver” per my Twitter output. Not clear currently how I would measure what I “take”. Ideally, I would want to get a sub-division of this into a more detailed split of what my tweets have related to.
6. The 11% for Links to Music etc seemed high and these are mainly for my benefit re what did I watch on TV, what podcasts did I play etc and hopefully also provides people with some background to me as a person re what content am I consuming directly myself. I do not know how much of these are then used by others.
7. You will see in the tweet above with the stats, I am keen to take any feedback, comments, questions, suggestions etc on what I did and how plus any ways of doing this more efficiently and effectively.
8. And then I wondered how my 2016 tweet output numerically compared to previous years which resulted in these figures:-