From Social Data to Social
Insight
Our perspective on Social
Media Research
From Social Data to Social Insight
Agenda
Our kind of social data
Social insight – what it is and how we get to it
Sharing the love
Who? Where? Why?
What next?
Q&A
What we mean by social data?
And what are the limitations…
‘...Informationgently but but relentlessly drizzlesus
‘...Information
gently relentlessly drizzles down on down on
in an invisible, impalpable electric rain’.
us in an invisible, impalpable electric rain’.
Hans Christian von Baeyer
Hans Christian von Baeyer
4
When people ‘research’ the social web, they usual get no further than monitoring.
Data on top of data. It’s useful, but it’s not insight…
500
£
0
Monitoring helps you aggregate relevant content but it doesn’t help you
understand that nature of the conversation…
So what is social insight?
And how do we get to it…
Why is an insight is like a refrigerator?
11 Images courtesy of Magnetbox from Flickr
Once you look into it, the light comes on.
Jeremy Bullmore
12 Image courtesy of crazytales562 from Flickr
Social insight is simply an insight derived from social content. The same rules
that apply in traditional research apply here. Otherwise it’s just a badlands
Image courtesy of Orin Zebest from Flickr
When undertaking social media research, it is vital to know what you are
assuming...
“There are known knowns; there are things we know we know.
We also know there are known unknowns; that is to say we know there are some things we do not know.
But then there are also unknown unknowns – there are things we do not know we don’t know.”
Donald Rumsfeld, 2002
As questions affect answers, so search terms affect the content you use to draw
your insight...
Image courtesy of Sean MacEntee from Flickr
just beware of face value
Image courtesy of tuppus from Flickr
And don’t read in what just isn’t there
Image courtesy of Andre Charland from Flickr
From Big Data comes Big Insight
Purchasing history
CRM Social CRM
and digital channels
Traditional research Social media
Multi-channel
?
behaviours
Real-time
experience
Big opportunities – comes from aligning insights
Source A Insight
Source B Insight
Source C Insight
Source D Insight
Source E Insight
Source F Insight
Source G Insight
‘Big Insight’
So the obvious cry is less data more insight. But before getting into that, I want to spend a bit of time looking at how that second layer of analytical data is generated because it is a crucial first step in any social media research. We just feel it isn’t the ultimate destination for social media research – you need a blend of different tools to interrogate the raw data and you need the right people to craft the insights...
So the most salient and widely used tools for scooping up all this social data are ones we’re probably all quite familiar with. They provide various statistical expressions about the volumes of conversation regarding your brand. Like any new source of information the focus is on getting the data and getting data representative of the social web, which is no mean feat by the way. The tools that are available range from the consultative and the DIY and on the whole they’re incredibly powerful tools for scurrying out across the social web, scanning millions of pieces of content and then giving you an array of statistics about that content in relation to your brand.
But when we talk about social data, we’re referring to the raw content that is created by people as well as the content about people. Multi-media content across forums, networks, microblogs and blogs. Content which, for any one study, is usually counted in the thousands and millions. This is the reservoir of opinions, attitudes, perceptions and behaviours from which all analytical data or insights can be drawn. And due to the sheer volume of data you can generate from that - the vast majority of social media research focuses on aggregation and metrics, which is almost always a quantitative dataset, and offers that at ‘insight’ Which it isn’t. It’s data on top of data.
The volume trend over time chart has become an iconic representation of the kind of data generated by a social listening tool. And while at this stage it is just data, it gives you an overview of how your brand is perceived across the social web, where it is talked about and who is talking about it and analysed properly social data can be a crucial foundation to developing your social media, PR or communications strategies.
At this juncture, it would be remiss of me not to mention a particular bugbear with monitoring – automated sentiment. As far as we’re concerned pure automated sentiment really doesn’t work. In our experience, whilst you can derive headline figures from it, these might – at best – be 60% accurate, so slightly better than the toss of a coin.
It is possible to conduct some more meaningful sentiment analysis, though it has to be humans doing the leg work... you have to be social and to get social to do the research - without that contextual understanding there's a real danger you draw the wrong conclusions. The need to understand who, how and why people use social media is absolutely critical – to account for things like venue effect (cite white paper from David Schweidel, Wendy Moe and Chris Boudreaux from Wisconsin School of Business) within the social web and to be able to able to actually draw actionable insights from the data. We use a blend of incredibly powerful tools such as Autonomy and Crimson Hexagon and have the luxury of a large team of analysts to draw on for specific work.
Let’s start off by considering what an insight is. It’s a word that is frequently bandied about. It has a number of different definitions. The best I know is this one…’Why is an insight like a refridgerator
… Once you look into it, the light comes on. We would argue that a social insight is just the same. It does the same thing. It should and can perform the same role.
If a social insight is simply an insight derived from social content, then many of the same qualifiers and watch-outs that apply to other methods of insight derivation also apply to the process of getting to social insight - the main differences are around how those qualifiers and watch-outs need to be applied in relation to social content. But the first thing to remember when seeking to get to social insights is that these qualifiers and watch-outs are necessary. They do apply. Otherwise, the process of social media insight identification will will gain a deserved ‘Wild West’ reputation. So, what re the key tenets to bear in mind when seeking to get to social insights?
Donald had a point. When undertaking social media research, it is vital to know what you are assuming. It is useful to have a hypothesis. But it’s very important that the assumptions don’t lead the research – especially given that – unlike consumers in a focus group – content can’t argue back, it can only quietly present its case and hope that you, the reader, are sensitive to its nuances and open to having your own mind changed.
Similarly, questions affect answers. A good researcher will think carefully about how to frame a question. Just because, when getting to social insight, you don’t actually ask questions direct to consumers doesn’t mean you can avoid taking the same care with your ‘questions’ – your search terms. A narrowly defined set of terms can be appropriate, but either a very specific brief or background research are necessary in order to arrive at these searches. And if the content returned gives you any cause to question the searches you have used, any hint of a hidden story, you must change or broaden the terms in order to test the approach and unearth the story. A broad approach e.g. just searching with a brand name is more likely to reveal those ‘unknown unknowns’.
Thirdly, beware of face value. Insight derivation based on social media is based on what people write – the opinions they express, the claims they make. Stated behaviour. Stated attitudes. This is actually no different to any other method of research, but it applies just as much if not more so for social insight work. Why might someone be expressing what they are expressing within that particular type of social media? At an aggregated level, individual motivations arguably matter less, but it is worth considering whether a particular motivation or agenda might be sitting within that aggregated data. Cross-referencing with other sources of insight helps here, but what can also be helpful is to analyse opinion by social media source. Twitter content, say, has different characteristics and arguably also motivations, compared to, say, forum content. This is obvious – but really important. We looked last year at the response within social media to the launch of Sky Go. We found a lot of balanced content within forums. Turned out that much of this resided in Sky’s own customer service forum – where customers were going to try to resolve problems but – being engaged enough with Sky to know about its forum – were prone to offset their experiences with comments about the positives of being a Sky customer.
But equally, it is easy to read what you want to read into content. You can draw inferences as to needs and motivations, especially by considering the context of the source, the wider exchange the protagonist is engaged within, and the needs and motivations espoused by others. But without the ability to ask and check, or the signs that face-to-face can sometimes can suggest that what someone is saying is not all they are thinking, unchecked ‘interpretation’ can run the risk of reading into content what just isn’t there.
Suffice to say that as for all research, but for social media research especially, context really does matter. It shapes our understanding of the content.
And, as with any other research, the more sources, the better – more accurate, more certain – the result. Social media research, producing, as it does, actionable insights, can be used in place of the focus group or the survey. But, it is better used together with other forms of research – desk, quantitative, qualitative, and can be used to frame, parallel or follow-up on other research sources and methods.
So there is a more strategic way of exploring and analysing the raw social data. Let’s take a look at some examples of social insights and how they differ from the data that might be generated by a more typical monitoring or listening tool...
A great place to start is BT – who’ve been leading the adoption and integration of social technology in the UK for years. What I like about this, and why it’s a good place to start, is that it’s an intelligent re-appropriation of the more traditional monitoring data and something that should be a lot more common. They used Sysomos to identify social media mavens and influencers, which is a relatively straight forward thing to do, but rather than use that as validation for any existing activity they invited those guys into a specific MROC as a crowdsourcing community. While there is no human layer to it in so much as these ‘mavens’ are identified by the algorithms in the technology – I think this is mitigated by the fact that they’re inviting these guys into a more collabrative, conversational environment which is very ‘human’.
This is a slightly deeper social insight that was part of a study we undertook last year looking at consumer attitudes to 5 of biggest handset brands in the UK along with the Android brand - I think this is a great example of conducting social media research that actually identifies potential opportunities for a brand– a blend of (human) sentiment tracking, conversation theming and share of voice around the top 5 mobile handset brands and the Android brand. Looking at it this way, we find that Android is the biggest competitor to iPhone in terms of engagement. Nokia, whilst it does have a warm, fuzzy feeling around it, is very much seen through a nostalgic lens currently which could be a powerful asset. Samsung underperforms its market share in terms of share of the conversation, and is not yet making the inroads as a brand within social conversations that its commercial performance might suggest it should be. Apple advocates clearly enjoy being part of a community that seems to have the upper hand and regularly remind Android users. Apple’s closed system and lack of compatibility with other platforms is seen as an advantage by loyal users and helps retain the notion of exclusivity. This perception of exclusivity is core to the appeal of the Apple franchise. In contrast, Android fans are enjoying being part of a community supporting a challenger brand which they see as having equal capabilities. They see Apple as representing the status quo. Openness and compatibility is a key pull for the Android platform and gives the impression to users that the system is superior. So what does that mean? For Android, painting Apple as representing the status quo and ‘following the herd’ could help reinforce their position among a wider audience.
M&S posed several questions on their facebook forum about new products and flavour combinations, a lot of which received several hundred comments. They then used a free facebook analytics tool and a lot of man-hours to read through to comments to find threads and trends. One theme that kept reoccuring from several different product conversations was to ‘giant-size’ sweets and chocolates, in particular the walnut whip. So – in the spirit of ‘social’ this was taken and developed very quickly and we can see them in the stores now and has gone down very well. What I like about this was that it isn’t a co-creation project in the classic sense of the word – they didn’t post the direct question about walnut whip – they analysed the comments and found a pattern. Perhaps crowdsourcing in it’s truest sense?
Conversational themeing can have a uncover a multitude of valuable insights and one way of using this approach to social data that almost always goes down well with clients is mapping a brands customer pledges and brand values against conversation content. This is one we did last year around Asda’s brand pledges. What is interesting is that all of Asda’s customer pledges, which are the cornerstones of their brand messaging, is resonating people across the social web. There are also two elements to how and why this is interesting – firstly that they’re on the right track with communication their brand values but also how this is reflected in the in-store experience. There is a strong tendency for consumers to report their experiences as they shop or immediately afterwards, those first moment of truths being expressed on social media is a hugely powerful insight for retailers. So not only can this identify areas of a brand promise that might need specific attention , it can help gain early warnings of any changes in performance and indicate opportunities to innovate the in-store experience.
Crisis management is probably the most common use of social media insights in businesses at the moment. Quite simply something has gone wrong or something negative has been published about your brand and it’s all kicking off on your social media feeds. Twitter storms being the worse example of crisis – although as an aside, we recently discovered looking at a number of recent Twitter storms that, in many cases, the traditional media include the public’s response to the news by looking at Twitter as a means by which to extend the story and to also create a new angle; the BlackBerry outage in October 2011 was a perfect example of a self propagating Twitter storm. Many brands not only work hard to avoid the news from affecting their brand reputation long term but also use it to engage their followers, fans or customers further. We recently heard of an example cited by the BBC talking about how they had handled their website outage last March – by analysing conversations, and understanding the key themes, they could adopt a cohesive approach to messaging their followers about the crisis and at the same time drive further engagement with the brand. Their approach meant that by using several social media channels they could reassure their audience as well as acknowledge the impact of the crisis. Virgin Atlantic also used the insights derived from social media feeds to drive their communication around the pilot strike threat in June 2011. Their main objective was to ensure that the news shared alleviated customer concerns around the potential impact for them. Generally this is the remit of the PR, Media relations, Social Media relations or even Customer Service teams but you could argue that the responsibility for developing insight should lie with the insight team, but the whole process should be handled by a cross-functional team.
Using social insights to inform your Social media strategy and content seems to be a fairly obvious application, although this seems to remain largely unexplored by brands. It seems that brand are sticking to the tactical use at the moment (i.e. Crisis management) and are not using social insights to drive their content. Social media content broadcast by brand seems to be largely driven by their wanting to leverage engagement by either entertaining – O2 and Sony Eriksson are key examples of that, taking on an advisory role or simply engaging for the sake of engagement (e.g. KitKat Chunky Campaign). There is a case to be made for analysing the content of spontaneous brand engagement with your brand to help formulate how you want to engage your audience via social media and through which social media channels. This largely sits with the social media team although this should be integrated into the brands overall marketing plan as well as the integration with overall insight. The analysis of previous social media campaigns should also feed into this of course.
Social Insights are definitely something which are left largely unexploited when it comes to ongoing brand reputation tracking and measurement. Measuring the effects of a specific event or crisis around a brand is more common, but the systematic tracking of conversations against values and image constructs is rare. The idea is not a far stretch though as brand reputation is often measured by PR agencies through the analysis of editorial content and to ensure that their PR campaign efforts are on track. Doing the same using social media is a next logical step. Brand image ownership normally sits with the Marketing team and brand tracking is something that is normally managed by the insight team.
We have talked about crisis management and twitter storms a little earlier. But social media can be used beyond what we would call customer service management Another area in which social data can bring insight is in the area of customer experience. We recently did some work to understand whether Asda’s 5 customer pledges were reflected in customer conversations based on their recent in-store experiences and more to the point were being met when customers visit store. What we found by looking at conversation content around these pledges is that they were discussed in almost half of all the conversations mentioning Asda, proving that you can monitor customer experience using social insights. For retailers, customer experience is key and being able to track whether the key elements of the customer experience are delivered in real time, can help gain early warnings of any changes in performance, and indicate opportunities for improvement. The other application to identify what drives positive customer experience which is something that we applied in a project we conducted recently around customer churn for mobile networks. You could argue that this is the remit of customer services, but customer experience management and development reaches into every team within the organisation and while we are not suggesting that they are involved in developing the insights, they need to be championing the process of collecting them.
The last area is innovation and research & development. There is a mis-conception around social media content which sees the conversations being taken at face value. Analysed with the right lens and the right depth, saocial insights can be used to understand consumer/customer needs. Debbie Weinstein Senior Director, Social Media Innovation, at Unilever was heard last week at the Social Brands conference saying: "Now we are using social media to feed into innovation and research and development“, indicating a shift in the way social media insights are being used. There has also been a recent shift in the way that social media is looked at specifically in terms of ROI. Sienne Veit, Head of New Technology Business development was recently heard saying that the ‘real value of social media is insight’, which points to large organisations integrating social media as a source of consumer research and insights.
Our suggestion is that getting to ‘big insight’ will require an approach based on lining up the insights provided by each of the sources, and looking for the golden thread that runs through them – an aggregated insight that is more than the sum of its social (and non-social) parts.
For the better part of two decades, businesses have been looking well beyond traditional research – quant and qual – for data and insight. Companies are increasingly taking data on board from social CRM and their own digital activities, and from wider social media – which is where we currently come in. All the while, of course, traditional research has been innovating. We’ve had the four hour super-consumer group, ‘creative consumers’, and of course, the huge changes brought by online research. Online focus groups, communities etc. And that innovation is being paralleled within every spoke of the wheel. And there’ll be more – as consumer behaviours increasingly span multi-channel experiences, understanding them within each of those channels and joining up the dots will become increasingly important. Data from real time customer experiences – call centre data, for example, is another obvious next step. And then, there will doubtless be myriad new sources of data and insight, some of we haven’t worked out how to mine yet, others which haven’t even been invented yet. There’s a great opportunity to identify ever more insights as a result of all this, but there’s a danger too – that we go full circle, that the drizzle of data we spoke about earlier becomes a monsoon. As all of these sources of data stack up, even if they are all being mined for insight according to the tenets we established earlier, the multiple insights that might be extracted from these multiple sources demand a different approach to the final process of getting to – in the spirit of ‘big data’, let’s call it ‘big insight’.
Our suggestion is that getting to ‘big insight’ will require an approach based on lining up the insights provided by each of the sources, and looking for the golden thread that runs through them – an aggregated insight that is more than the sum of its social (and non-social) parts.