After lunch we jumped into our second thread of the day, Second Dimension: Social Media Research and Online Communities. Session chair Charles Hageman started with the question, what is social media research? How do we define it? Text analytics, network analytics, transient and even confusing were all shouted from the stands. It’s clear there are many facets of social media research, but Charles groups it all as a “tool”. Hageman went on to define MROCs, communities used for market research, again they can take many forms, but the link between them all is that they involve communication. It’s people talking to each other, and they don’t run on their own. With that the session kicked off with Socialising Research presented Michael Rodenburgh of Ipsos Open Thinking Exchange in Canada.

Michael first started with the new normal, a society where information flows freely between brands and consumers, where social media has created heightened consumer expectations for brands to engage. This is having a profound impact on how we research. The old methods are becoming irrelevant. He went on talk about how Ipsos are using “socialised research”, blending advancements in tech, and using the cultural shift towards social interaction and brand interaction with consumers. Using Active, passive and interactive techniques. Facebook has 1 billion users, the single biggest of consumer panel, as Michael stated “It contains pictures that tell a story about my life.”  It can be a goldmine of data.

Michael then talked about a recent study carried out by Ipsos looking at how to harness Facebook data to create insights in TV show audience size and social activity. Using an API tool on Facebook, they connected with a group of their panel-survey respondents. They found that those who shared their FB information with Ipsos were almost demographically identical to those who refused and there were no differences in the tech-savviness of each household. They also found a pretty significant correlation to audience size and social engagement. A non-linear relationship, an online double jeopardy effect, they found larger shows draw disproportionately more social engagement than small shows. And that shows that out-perform the average all have hosts with a high emotional connection with the audience.

So what can we conclude about FB data? Well according to Michael it can allow TV shows to leverage the affinity of over-indexed shows that share fans on Facebook, allowing them to direct on which shows to advertise and what consumer products can be approached for advertising. But as we’ve heard already today, such bid data sets have issues. With coding and tabulation and the visualisation of the data being two prominent issues. So if we are too use data from platforms such as Facebook, we need to look at several factors:

  • Technology – purpose built to handle big data
  • Training – Data scientists are not data analysts
  • Organisational structure that supports it and manages client service delivery.

Up next on this afternoon’s thread was Martin Einhorn of Porsche and Olaf Hifmann of Skopos with Porsche Research Rally. They asked whether social media research can replace traditional methods. Porsche started a study to identify all the possible touch-points consumers have with the brand. They started using the traditional methods of online questionnaires and focus groups while starting a social media in parallel to accurately compare and gauge the information collected.

They found that 90% of usable data came from forums rather than Facebook and Twitter. They also found they had a huge amount of data, they managed to cut some of the wheat from chaff using software, but found they had to then examine everything manually. They found that a deeper use of software was incompatible with an accurate segmentation of the data. When analysed they found the insights from social media were more general and although they provided clear and valuable customer stories and additional ideas, there was little context and motivation in posting on a forum is slightly biased, good experiences led to more comments.

Olaf thought that although social media data is powerful, it has limitations and needs to be coupled with traditional methods. There is also a huge problem is software, depending on which software you use you will get different data and different insights.

For the final presentation of the first afternoon thread we had Joseph Blechman of AOL in the USA on his paper, AOL Breaking Social Media News, which looked at uncovering the hidden trigger of user generated content engagement. Joseph started with the current state of play; most research focuses on authorship, counting comments and sentiment analysis. Bu this is missing half the equation. Most people read rather than write comments, and focusing on authorship misses the reader experience.

AOL wanted to know how user generated content effects and contributes to the reader experience. Using a sample of some new sites with the most vibrant communities they read 100s of comments and analysed the DNA of fact based comments on these platforms. They looked at factors such as whether they commenter had an icon/avatar, whether they used their own name, whether the comment praises or critiques the original comment and clarity of comment. From that they created a survey where respondents self-selected an article of interest and went on to rate the comments over 9 dimensions on a 5 point scale. With two of the key metrics being “makes me want to join the conversation” and “makes me want to read more comments”. They found that personal identity and community involvement of the commentator is important but the relationship of the comment to the original content is most important. Prior to the study AOL assumed that individual substance would be the most important driving factor to read more comments, but in the end it was not.

Joseph went on to explain that this study gave AOL the ability to score individual comments and compare to others, so they studies and hand-coded 3200 comments on a draft of AOL sites and found:

  • Readers turn to comments for additional information and perspectives
  • Readers like to associate a human to the content
  • And Facts are less important than reaction.

This can be used on a broader lever to diagnose community health, benchmark against competitors and what users and actions should be encouraged within the community.

After a short coffee break for the ESOMAR delegates we headed straight back for part II of the Second Dimension thread and a continued look at MROCs and social media. Charles Hageman kicked us off with the presentation Solid as an MROC with Istvan Hajnal of iVOX in Belgium. Istvan was here to talk about the use of text analytics in research communities, kicking off with the usual background stats he claimed that 36% of consumers are willing to participle in a branded research community (InSites) and that 64% of research providers or suppliers are planning to use online communities in 2012 (Greenbooks).

But Istvan wanted to talk about the role of the moderator and researcher in MROCs rather than the respondent. Respondents have changed, they are completely mobile, they answer surveys through games and can access any community they want to. But what about the analyst and what about the clients. Well they haven’t changed. The analyst is doing the same work and the client still want it cheaper, faster and better. But worryingly we are seeing more and more clients are making decisions based on gut instinct rather than evidence.

iVOX created an experiment that focused on the researcher side, rather than respondent. Creating 4 communities, 2 with additional tools and 2 without. They used a number of prototype tools to help the researcher. Including moderation tools, such as a no-show dashboard and reaction and response tools. They also created analysis tools which looked at text selection and latent semantic analysis. These text analytics tools provided a starting point, though only a starting point as text analytics is far from perfect.

Istvan went on to say that not only did these tools make the process more efficient from the researcher/moderator side but also saw an increase in number of participants, a decrease in drop out rates and increase in content generated by users. Though he made it clear in his final point that although these tools should be explored more, automation is not always the solution. The personal connection between the moderator and respondent is still very important.

To follow we were joined by Rakesh Kumar of Firefly Millward Brown in Singapore and Madhumita Chakraborty of PepsiCo in India who presented DigiAsia, which looked at how social media was playing a role in the changing youth makeup in Asia.

Affluence, education and connectivity is creating a massive shift in Asian psyche, currently manifesting it’s self in changes to youth culture. Madhumita went on to state there are 1.3 billion online users in Asia and the Middle East. But how much does social media play in these changes in youth culture? After all, people are social by design. If you are connected all the time it’s bound to have an affect on how you interact in with the world.

Millward Brown and PepsiCo approached this study with a number of hypothesis. Does social media represent a new dimension of youth identity. Do offline personas differ with online personas, are online personas edgier? Do online personas remain constant across platforms? Does social media allow youth to experiment more? The study took a group of youths (16 -24), recruited both online and offline. The team shadowed them online for a period of a month and during the period they also carried out face to face meetings, before and after social media shadowing study.

They found that the social media identity is not a new dimension, but like the offline persona it adds an aspect to it. They also found that although social media does offer a safe haven for experimentation, it only offers a temporary escape from my regular life. They also found that the online and offline personas might merge as youth pursue their passions online but it may also be different. What they found was that the audience is the object of the conversation, gaining followers is very important.

In the last session of the day, Dan Brilot of YouGov and Anita Penn of the BBC in the UK joined us for their presentation The BBC Olympic Flame, looking at understanding social media activity during the Olympic Games and whether the use of social media drove coverage of the games. Dan started by stating the unofficial theme of today’s presentations; there is some good work being down in social media research but the technological capabilities and research methodologies have not caught up yet. He went on to say that although half of all companies in the UK claim to have a social media strategy,  many didn’t know what it was, who was responsible for it and the KPIs being used to measure it.

Dan harked back to one our previous presentations, saying that social media can offer a wealth of insights, but it does have to be done responsibly, the fundamentals and the science is important. He went on talk about the study carried out by YouGov and the BBC. They split two groups of respondents, all active in social media into two groups; followers and non-followers of the BBC, then went to an number of lengths to ensure they were demographically identical. Using test and control groups, pre and post games. They were able to record mentions on social media and respond directly with a survey.

So what did the study say about the influence of social media on watching behaviour? Anita Penn told us that the BBC followers claimed to watch more and differences in some pre and post testing showed an inference of social media as a factor for increased coverage views. It also showed that word of mouth was the biggest factor on influencing behaviour. She went on to say the study can be used to strengthen the BBC brand, personal and emotive statements saw uplifts between both BBC and non-BBC followers. Social media can be an important was in which we connect with audiences and strengthen the relationship they have with you. But although there seemed to be an uplift in viewing figures based on social media, there is still along way to go before we crack the answer for sure.