Thursday, 9 February 2012

Durham Brewery White Stout - analysis of twitter launch

In my last post I previewed an event Durham Brewery had come up with to launch their new beer, White Stout, both in pubs and on twitter at 8:30pm on the 8th February. The idea was to get as many people as possible drinking and sharing opinions on White Stout. This post analyses how successful that was.

This post won't discuss the beer itself. If you'd like to read about that, Barl Fire has written a great post about it which you can read here. Also well worth a read is Phil Hardy's post about the fantastic beer and cheese matching event he set up on the evening here.

If you've read my previous posts, you'll know I like numbers and analysing data. If it can be munged into an excel pivot table, I'm game! After the event, I greedily snaffled a week's worth of tweets mentioning White Stout, or using the #whitestout hashtag and set about trying to make sense of it all. For the curious out there, I did that with the Windows version of Archivist, which lets you search using the twitter API and spits the results out in a tab delimited text file, which excel will suck in with a few clicks.

The analysis below covers 1332 tweets which were sent between 08:52 on 2nd February and 22:00 on 8th February.  The below chart shows the distribution of those tweets by date (note: you can click on any of the images below to see them full size):


So the above tells us, unsurprisingly, that the bulk of those 1344 tweets (983 to be exact) were sent on the 8th. Carrying this forward to midnight shows there were actually more than 1000 tweets sent on the 8th. That's a lot of tweets. So let's break those down by hour. The below graph shows how these 983 tweets were distributed between midday and 10pm:


For clarity, 20 on the x-axis above covers the period between 20:00 and 20:59. So there was a huge spike in that hour - which is where our launch time of 20:30 lay. Let's look more closely at that. The below chart shows the number of tweets per minute during that hour:


So the volume of tweets actually peaked at 20:39, where there was one sent just about every two seconds!  Who was sending them? The below pie chart shows the distribution by username throughout the whole period, with the top 20 odd tweeters called out by @name. It's followed by a list of users with the number of tweets sent total:


 

Looking at the above, the top 10 users sent a total of 664 tweets to their cumulative count of 5115 followers. Now, consider that a total of 162 twitter users took part and you start to get a feel for the reach of those tweets - probably well in excess of 10,000 people saw a #whitestout tweet! In summary, this event was a massive success and I believe #whitestout could well have trended if a certain England manager hadn't resigned an hour or so before the event kicked off.


Now onto the fun part. I used wordle to create some word clouds which show the words used most often in those tweets. The first one below covers all the tweets and isn't filtered, other than removing the tag itself:



The second cloud below, to me is the most fascinating part of this analysis. It takes the descriptive words used in tweets around the time people were opening and tasting the beer and aggregates them, again, the larger the word the more frequently it was used. To me this is like the 100+ people all talking loudly in the virtual #whitestout pub, sharing their views on the beer - the very essence of what the event was trying to achieve:


So there you have it, a number and word crunching view of the how the night went down. If you took part, what did you think? Is this an effective way of using social media to promote beer and engage with consumers? If you're a brewer, would you consider doing the same or something similar in future?

No comments:

Post a comment