Author: Laurence Borel

In or out? A digital analysis of #brexit

Before we deep dive into the data, I’d like to clarify this post is solely intended as a quick analysis of the volume of conversations sentiment and traffic surrounding a vote that could significantly change the world we live in forever. This analysis looks at social media and traffic data from both the Remain and Leave campaigns.

Social Media data were collected during 1st October 2015 and 31st May 2016, when both the #voteleave and #strongerin hashtags started emerging.

Stronger In:

  • 208,943 mentions were recorded during period and peaked in May 2016
  • As we can see from the word cloud below, the key arguments posited by the camp appear to be around trade agreements, businesses, universities.
  • We can also see that those from the Brexit camp often intervene within these conversations, as demonstrated by the prominent #voteleave hashtag within the word cloud.

Stronger in volume and sentiment

Stronger in Word Cloud

In terms of digital channels. Stronger In are being smart with their PPC stategy. This is what happens when you Google ‘Leave EU’…

Stronger IN


  • Conversely, the #voteleave hashtag received 268K mentions during the same period, and appears to be following a very similar curve in terms of mentions as the #strongerin campaign
  • Key topics of conversations tend to be around taking control and stopping the EU/euro-scepticism, priorities and Brussels, but no other clearly identifiable themes of conversations emerge.
  • The mentions around the #voteleave movement peaked in May, with The Ordinary Man Twitter user, driving a large volume of RTs throughout the month

#voteleave sentiment and volume over time

#voteleave word cloud

Traffic analysis (April – June 2016)

Of course, the fact that there are multiple segmented movements for the Leave campaign does not help with their campaign traffic-wise… Stronger In however beats with the support of a PPC campaign, which appears to have kicked off in April… Because the traffic significantly dropped in May, I would suspect that they have reduced their PPC investment significantly, with the aim of potentially increasing spend again in the final few weeks of the campaign.

Stronger in vs. leave

Scrunity of the Stronger In vs. Leave Take Control however depicts a different story… Vote Leave had a clear lead traffic-wise until April 2016, when the Remain camp kicked off their PPC campaign… As of May 2016 with significantly reduced levels of paid search investment, both camps are head to head. Remain vs. vote leave take control

Final stats as of 1st June 2016:

  • Remain traffic: 363,112 vs. Leave 372,912
  • Remain social mentions: 208,943 mentions vs. Leave: 268K
  • Remain social media count: 444,703 likes, 32,285 followers. vs. Leave social media count: 439,356 likes, 48,785 followers.

If we are to go by these numbers, the Leave camp marginally wins…

On ‘Digital’ Narcissism

Narcissus was the son of River God Cephisus and nymph Lyriope. Narcissus’s mother was told by the blind seer Tiresias that her son would have a long life, provided he never recognized himself. However, his rejection of the love of the nymph Echo, drew upon him the vengeance of the gods. As punishment, Narcissus fell in love with his own reflection in the waters of a spring and pined away. As for the nymph Echo, the loss of her love made her fade away until all that was left of her was her voice. 

It is time for us all to stop misusing (and over-using) the term narcissism in the context of social media use such as selfies. The sheer volume of articles in relation to ‘selfies narcissism’  returns 268K results, whilst digital narcissism, a construct yet to be defined by academics and psychologists alike, returns a staggering 502K results. Let’s stop this unfounded click-bait driven media panic, shall we?

Let’s start by defining what narcissism is… 

Havelock Ellis (1898)
 first  used the term Narcissus-like to refer to “a tendency for the sexual emotions to be lost and almost entirely absorbed in self-admiration” (Ellis, 1898, p. 890).
narcissus-caravaggio-300x363Freud’s ‘On Narcissism’ (1914) is widely considered to be an introduction of the author’s clinical theories. Freud describes narcissism as an excessive degree of self-esteem or self-involvement. Central to Freud’s theories are problems of the relations between the ego and external objects. Freud (1914) drew new distinctions between ‘ego-libido’ (investing the energy of libido into ego, the equivalent to “having sexual desire towards (my) self”, and ‘object-libido’, the idealisation of an object, which becomes aggrandised and exalted in the subject’s mind.

But to keep things simple, let’s simply refer to narcissism as “excessive self-love or vanity, self-admiration, self-centeredness” (Oxford English Dictionary, 2014).

So yes, when looking at selfies, one might think on the surface that selfies are a narcissist act. But are they really? What about No Make Up selfies? Are these selfies really narcissistic, or are they, on the other hand a form of communication or perhaps political empowerment?

Measuring narcissism 

Another issue which I have, particularly in regards to the media and their click-bait strategies, is the fact that narcissism, similarly to the Big Five personality traits,  can be measured. Few of these click-bait articles have empirical evidence to back up their claims. Narcissism is, and has been measured with Raskin & Hall’s (1981) 40-item Narcissistic Personality Inventory (also known as the NPI-40) for decades. Whilst recent academic studies linking the selfie phenomenon to narcissism have used the NPI-40 as the main measurement the construct, I firmly believe that this measurement is inaccurate. Just because someone happens to be sharing selfies online, does not make them a narcissist. Correlation does not imply causation, although frequency of sharing selfies has been linked to higher degrees of narcissism.

Whilst recent academic studies linking the selfie phenomenon to narcissism have used the NPI-40 as the main measurement the construct, I firmly believe that this measurement is inaccurate. Just because someone happens to be sharing selfies online, does not make them a narcissist. Correlation does not imply causation, although frequency of sharing selfies has been linked to higher degrees of narcissism. Furthermore, as noted by  Chou and Farn (2015), the plausibility of applying such (antiquated) measurements to the Internet world still lacks systematic proof. What we need is a robust scale to define and measure what Digital Narcissism is and what its dimensions are. Cyberpsychologists, over to you…

Note: I have tried to keep this blog post short and sweet. I hope that this article will ignite a debate in the community. This article is by no means intended to be a thorough literature review. It wouldn’t fit on one page as we all know! 😉 

Photo via


Beyond positive, negative and neutral sentiment analysis

There has recently been quite a bit of chatter of late around social media listening tools attempting to analyse consumer sentiment. As we all well know however, these social listening tools often fail to correctly categorise sentiment at basic level (i.e. positive, negative, neutral), and more importantly, also fail to capture the real depth of human sentiment and emotions behind social posts. Marketers are stats junkies, even though these stats are often meaningless, and fail to inform as to what the next steps of your engagement approach should be. As Dr Jillian Ney puts it, ‘social media isn’t quantitative data, it is qualitative data on a quantitative scale.’

Let me explain through a quick case study. A couple of days ago, Domino’s Pizza announced the launch of their ‘Zero Click App’ which lets customers order a pizza with no clicks whatsoever. A quick search of mentions of the app using Crimson Hexagon (a tool which I love by the way) returned the below sentiment and media visualisation.

Total volume and sentiment

Great! There are 5K mentions, 15% of which are positive and only 3% negative. But, so what? When drilling further down into the topic of the mentions around the app (see visualisation below), yet again, I get little more than a pretty looking conversation map.  Don’t get me wrong, these conversation maps can be useful at times…

Conversations drill down

Another area of concern for me, is that these tools often fail to capture volumes of Facebook mentions accurately. Worryingly, brand communities such as Facebook pages are the online place where your most engaged fans are (and not necessarily engaged digitally; think brand engagement). These fans are consuming your content, talking about your brand and even interacting with your brand. Whilst, Facebook insights do a good job of measuring the reach of your posts, the number of clicks/views of your posts, fan demographics etc., these ‘insights’ serve no other purpose than (dis)pleasing online acquisition managers.

What if there was a better tool to understand these comments. Well, now there is thanks to Magicrowd, an online platform for qualitative analysis of UGC… on steroids! 

Magicrowd is the brain-child of artist and entrepreneur Aalam Wassef (and his team of 9). Aalam’s professional experience (including a stint at the prestigious National Centre for Scientific Research (CNRS) in France) led to his interest in understanding and analysing audiences.

Whilst, social media is typically analysed quantitatively, Magicrowd analyses UGC qualitatively and address three key questions:

  1. What are consumer saying?
  2. How are they feeling?
  3. What drives them to express themselves and feel the way they feel?

Although currently limited to Facebook, Magicrowd are planning on expanding the platform to offer Twitter, Instagram and Tumblr UGC analysis in the near future.

Too good to be true, right? Back to our case study; let’s see what Domino’s fans have to say about the new app. Here I am analysing fan comments posted under Domino’s status update. Immediately, you can see the basic sentiment emerging from Magicrowd is quite different from the results I pulled from Crimson Hexagon.

Simply copy and paste the URL of the post you wish to analyse, and manually code 10 comments by following on-screen instructions.  Magicrowd will then return a full analysis of your fans comments based on the initial manual coding. These emotions are analysed based on a number of   psychological, sociological & anthropological theories including Maslow’s hierarchy of needs, Jungian archetypes and Robert Plutchik’s wheel of emotions. These theories are combined with  data-science, NLP and machine learning to deliver the results shown below.

Magicrowd sentiment analysis

Magicrowd emotions analysis

Aside behavioural differences in social networking sites use (i.e. Twitter is for quick sharing which may explain the high volume of neutral mentions, whilst Facebook is used to express in-depth opinions), Magicrowd also analyses motivations and personalities with the aim of profiling consumers. By profiling your engaged customers, you can calibrate your content and hit a chord with your audience. Here’s an example below.

Meet the audience

Audience profiling: the creatives

Magicrowd is now in Beta with premium features being released in the coming weeks. Check out Magicrowd here and make sure you follow them on Twitter.

How’s your PhD going? Are you finished yet?

The 1 year and 4 months into the PhD update…

The dreaded questions any PhD student undoubtedly hears on a regular basis: ‘So, how’s your PhD going? Are you finished yet?’

‘Yes, it’s going, it’s going’… ‘it’s going’ is a good thing by the way. Not ‘going’ would be worrying. As for the second question, I’m internally cursing at you. Please don’t ask. I’m not finished, and I don’t know when I’ll be finished…. probably sometime in 2018 (Please send your CV if you’d like to apply for the role of ‘rich husband who can help me fund my extravagant PhD lifestyle’ – job title TBC-just kidding).

Most PhDs I know have been mildly traumatised by the whole experience ordeal. Somehow, I am still relatively unscathed.

Inspired by Laurence, here’s a  summary  of my PhD so far. PhD updates are highlighted in blue, and milestones related to additional PhD activities (conferences, publications, university lecturing) in pink.

January 2015-May 2015: Yes, I’m officially a PhD student. I love my topic. Now let’s write the literature review. Hmmm careful with plagiarism. Which referencing software shall I use? I’ll try them until I find one I like. Oh shit. Referencing software has messed up my word doc. I won’t use a referencing software after all. And I now have to add my 150 references of or so manually back into my draft. Fail!

June 2015 – seminar at Aston University: Oooo first academic seminar. I’m going to mingle with lots other PhD students. This is fun. Hmmm everyone is smarter than I am though… hmmmm.

June 2015: MRes hashtag research paper submitted. I’m going to be a published author. Maybe…

July 2015 –  ICORIA conference: Yes, my first conference! I’m presenting my MRes hashtag paper. I’m an academic! Or may be not…  Jesus, there’s so much competition out there. Must work harder.

September 2015: After 9 months of writing my literature review and not finding a descent research gap, my supervisor steps in. Amen! I wasn’t far off from finding my research question. Or maybe I was. It doesn’t matter. I love my topic!

October 2015: MRes hashtag paper rejected by journal. Do not pass go, do not collect $200 – feeling deflated. 

November 2015: First draft of my research prospectus submitted. What do you mean I need to make amends. What amends?

November 2015: French university has asked me to lecture digital marketing in March 2016! Happy! Must prepare lectures… 

December 2015: Second draft of my research prospectus submitted. More amends needed? What? OK, I shall spend Christmas amending my draft. Dissertation amends have now become a bit of a Christmas tradition anyway…

January 2016: Third research prospectus draft submitted. More amends? *bangs head on desk*

January 2016: Book chapter on Digital Branding and Analytics accepted. Yeah! 

February 2016: Success! My supervisors are finally happy with research prospectus. First study begins… Content analysis here I come…

February 2016: Yeah! The hashtag paper has been submitted to another journal. We’re hoping for R&R* (*R&R stands for revise and resubmit in academia, not Rest and Relaxation. hmmmm).

March 2016: I’m in France, I’m a lecturer! Students like me, but oh boy, correcting these dissertations is hard work.  Gahhh

March – April 2016: God this content analysis is painful… why did I say I would content analyse 2000 pieces of UGC? Why?!


To be continued… 

Photo via