What Makes for Good Twitter Traffic?

We studied all of the Twitter traffic from parliamentary candidates in Wales in the 2010 General Election. We also looked at the metrics associated with their Twitter accounts (how many people they followed, how many people were following them) and roughly scored their performance to identify the people who were making most effective use of the medium. It was interesting to find that the qualitative analysis (our subjective analysis) corresponded reasonably well with the crude quantitative scoring.

Of the candidates in Wales, perhaps only a dozen seemed to really understand the medium. Of these René Kinzett (Swansea West) was the most engaged, but there were also strong performances from Heledd Fychan (Montgomery), Matt Smith (Blaenau Gwent) and John Dixon (Cardiff North). As we noted in an earlier blog, successful Twittering did not always translate into an increased share of the vote – something which affected Dixon as well as Caryl Wyn Jones in Vale of Clwyd and Chris Bryant in Rhondda. However, Merthyr's Amy Kitcher performed extremely well in the votes and ran an exemplary Twitter campaign. But as one candidate reminded us, “Twitter was fun: the real campaign was elsewhere”.

So what made for a successful Twitter campaign? The critical success factors, we argue, can be summed up under the acronym SPREAD. This stands for 1) storytelling; 2) personality; 3) rich content; 4) excitement; 5) added value; 6) dialogue.

Twitter is a social medium, and as with all forms of interpersonal communication storytelling plays a central role. Celebrities such as Stephen Fry have shown how the public will take to an extended narrative, even when the content at times appears mundane. Similarly, Sarah Brown, wife of the outgoing Prime Minster, had over a million people following her on Twitter (which is roughly five times the number of registered Labour Party members at the time of the election). Welsh candidates went out campaigning, of course, and they got stuck in traffic jams, bought fish and chips, took their dogs to the vet, missed Doctor Who on TV having forgotten to set the video recorder, had to explain to their grandmothers what a videoblog was – in short, lived ordinary lives. Successful Tweeters told the story of the campaign on a daily basis and highlighted the many episodes that touched them, made them laugh, angered them. They watched the Leaders' Debates on TV and shared their opinions. Their narratives, like soap operas, were both mundane and compelling. It made them human.

Related to this is the aspect of personality, as conveyed by social media. The conversations and dialogues enabled followers to get a sense of the politicians as people. The younger Plaid Cymru candidates were often forthright in their views: one said how she would not trust a Tory as far as she could throw one, and others were similarly (and ironically) disrespectful of their opponents. This aspect led to party embarrassment outside Wales where during the campaign both Labour and Conservative parties were forced to suspend candidates after they posted off-message comments on their Twitter pages.

Effective management of Twitter included the use of rich content, something which younger candidates with smartphones took to well. The most obvious example was the use of photographs taken during campaigning and uploaded contemporaneously. These pictures were not just used gratuitously, though: the more effective ones told a story or entertained their audience. One Plaid Cymru candidate uploaded a picture of what she described as Lib Dem sheep, due to the large yellow markings on their backs. The same person posted a closeup of her face to highlight her crooked fringe – with a wry comment that this might be what a balanced parliament looks like. Candidates used video blogs, podcasts and other multimedia devices effectively, as well as more mundanely using Twitter to link to relevant websites and news reports.

Twitter also allowed candidates to communicate the excitement and energy of the campaign by drawing attention to their busy schedules. Typically they were campaigning 10 to 12 hours a day, 6 days a week – or at least Twitter enabled them to give that impression. Live Tweets during key points in the campaign such as the Leaders' Debates and during controversies such as Brown's bigotgate embarrassment added to the sense that history was being made and that the candidates were sharing this with their audience.

As with storytelling, Twitter was used well by some candidates to add value to the campaign. There were three ways they did this, as follows: 1) by the immediacy of the way they reported the events (and here mobile devices such as the Blackberry and iPhone came into their own), with the Twitter feed acting something like rolling news; 2) providing advanced notice of future events (meetings, hustings, debates, etc.); 3) supplying privileged insights into the campaign based on inside sources (something existing MPs were unashamed about, particularly those that had access to government ministers).

Finally, and perhaps most importantly of all, Twitter enabled dialogue to take place. The most successful Tweeters engaged in debate, answered questions, posed questions of their own. Most candidates, though, were woeful in this respect. Their Twitter traffic, to the casual reader, would have showed their interest as being one way – talking TO an audience in the way a politician in the previous generation would have done with a microphone and loudspeaker. We randomly questioned candidates during the campaign, or commented on aspects of it: hardly any took the trouble to reply. Similarly, we 'followed' every candidate in Wales that was on Twitter: of these just 2-3 were interested enough to 'follow' us back. Social media is about conversations rather than dialogues: many candidates were skilled talkers but had not grasped the idea that Twitter enabled them to connect with voters, journalists, party activists, political opponents and other key stakeholders.


Comment on methods

In order crudely to quantify successful and unsuccessful Twitter activity we took three key measures during the General Election campaign: firstly the number of followers each candidate had; secondly the number of people following them; and thirdly the number of postings or Tweets made by the candidate. We assigned a numerical value to each criterion, ranging from 10 for the most active, 1 for the least active and zero if there was no activity at all. So for example someone with the most Tweets posted would score 10 on that scale, but might only score 2 if their audience was demonstrably small. We therefore had three measures as follows: 'Listeners' had high scores for the amount they posted and scored well if they were following a larger than average number of other Twitter users; 'Talkers' by contrast used the medium a good deal, and had comparatively large audiences; finally 'Engagers' had high scores for listening and for talking combined.

It can be seen that a candidate could achieve a score of 100 for talking, 100 for listening and therefore a maximum 200 overall for 'engagement' (which combined both scores). In our study two candiates, René Kinzett and Heledd Fechan scored maximum points. In practice we found that most heavy 'talkers' also tended to score strongly on 'listening'. There were a few exceptions, though, but the methodology enabled us to exclude candidates such as Lembitt Opik in Montgomery who seemed to give up on Twitter during the campaign despite having a very large 'following'. It also meant that candidates who were extremely vocal on Twitter but who had only a small audience were similarly downgraded in our scores.

Interestingly, the crude data scoring produced a ranking which broadly corresponded with the subjective (qualitative) analysis of the content of the campaign: the same candidates emerged from this as effective communicators as were highlighted by the scoring system.


Evolving Twitter protocols: the future

Already the exponential growth in iPhone and smartphone ownership is starting to change social media. For Twitter the mobile device is ideal, enabling users to post rich content in real time, to engage in dialogues (albeit in 140 characters) and to tell stories without having to be tied to a fixed broadband. Users are already doing this, and increasingly social media conversations here and on Facebook are being mediated through multimedia content.

It was encouraging that we came across no real evidence of party interference in their candidates' Tweets (although some were rather predictably partisan). One widely recognized requirement for social media is that messages should not attempt to sell or persuade, at least not overtly; social media instead needs to engage through the authenticity and spontaneity of the message.


12 May 2010



The study was carried out in Wales by Robin Croft (Principal Lecturer, University of Glamorgan Business School), and in Yorkshire by Dianne Dean (Lecturer, University of Hull Business School). The findings of this study and the research completed in the 2009 European Elections are due to be presented at a conference of the Political Marketing Association in Thessaloniki, Greece, in September.