Social Media Analytics Expert Interview Series: Part 3
What Is Social Media Analytics? - 4 Experts Debate
February 28, 2011
As a lead-up to the Social Media Analytics Summit, Text Analytics News has partnered with Useful Social Media to publish a series of expert interviews with top Social Media Analytics professionals.
In our third installment, we reached out to four experts to gain insights into what social media data really is and where it comes from. The Social Media Analytics Summit will be held in San Francisco this April.
OdinText is a patent pending text analytics platform developed by Anderson Analytics, the first firm to leverage NLP for the marketing research industry. The firms’ clients include companies in various industries including Facebook and Starwood Hotels. In 2007 they were the first firm to conduct analysis of a large social network provider (together with LinkedIn).
Principal, Social Target
Social Target helps clients plan and implement intelligence and analytics capabilities based on social media and other sources. Their independent research helps them maintain the market awareness to match each client's needs to the best available technology and research services, and their original thinking inspires new applications that aren't in anybody's Social Media 101 presentation.
Gnip believes the thoughts, conversations, and debates expressed in public social conversation have unlimited value and near limitless application, and that no one person or company can possibly comprehend the full importance or potential uses of this information. The company’s mission is to make this ever-expanding universe of social media data available via a consistent and reliable architecture so the world can realize the full potential of this amazing stream of information.
Kami Watson Huyse
1. When you think of “Social Media Data,” what do you think of first? Second?
Tom (OdinText): First I think about all the data that we are generally not getting. The rich juicy stuff such as the connections between people on Facebook and the comments on their walls, even discussions in LinkedIn groups etc.
My second thought is around some of the key publishers that are facilitating the generation of such valuable content: Twitter, Facebook, Tumblr, Wordpress, Disqus, etc.
Nathan (Social Target): I was trained by the early concerns about what people were saying in social media—only it wasn't called social media at the time. So the first thing I think of is the content people create: the blog posts, reviews, tweets, videos, and everything else they put out there for the world to see.
The second is the data created by all the observable actions people take in social media settings: connecting with friends; following interesting people; sharing, tagging, and linking to content; clicking on "favorite" and "like" buttons; voting on reviews... Most of the new features in social media create still more data, and the young companies that dominate the space don't seem to be out of ideas.
Kami (Zoetica Media): Social media data is at the heart of understanding your community. Far from being cold and impersonal, data can tell a story that intuition alone cannot deliver. As much as we like to believe that we fully understand our community, what people say and what people do are often very different. Data can help to guide intuition.
For that reason, the second thing I think of when I consider social media data is it's importance as a tool to diagnose, prioritize and evaluate what you are doing as an organization and use it to make course corrections.
2. Do you think there is currently a common understanding as to what constitutes social media data?
Nathan (Social Target): Definitely not. I can usually tell what a person's job is by the way they discuss social media data and what it's good for. That specialization is natural, because there's so much to learn, but it's keeping us from having a common vocabulary across disciplines.
Superficially, you can measure at least four categories of data in the social media context: the content people create, information about the creators of that content, information about the audience of the content, and information about how people connect and interact. I don't know if we'll ever agree on our terminology, but we can avoid misunderstandings by being clear about which data we're talking about.
Kami (Zoetica Media): I think there isn't a clear consensus. Is it text analysis, sentiment analytics, or something else? I also think it is being marketed to a constituency that doesn't necessarily have the background to understand some of the differences. There is also a difference between an academic view of data and harnessing data to provide actionable insights that can drive business decisions.
Tom (OdinText): Common to certain groups perhaps, but not between groups. Developers, for instance, often begin building a tool because a certain type of data is easy to get to. Klout is an excellent example. It’s totally reliant on Twitter even though less than 9% of the US population is on Twitter. Facebook is far more popular of course, but you can’t get to that data without users opting in.
But then if you look at groups who use some of these tools to make marketing decisions, they don’t understand what’s missing. Someone could be very influential in social media and not be on Twitter or Klout. If they have 5,000 connections on LinkedIn Klout won’t know, and the maximum shown on LinkedIn is “500+”. Just giving an example, I think Klout can be a useful tool as long as you know what you are seeing and what you are not seeing.
3. There’s a lot of social media monitoring vendors out there now. How do most of them get their data?
Tom (OdinText): Great question. As our text analytics platform OdinText is data agnostic, i.e. our clients are just as likely to use it for analyzing their Net Promoter Score (NPS) surveys or CRM/Call Center logs as with social media data, I’m not afraid to answer that question honestly.
What a lot of firms who specialize in just social media monitoring won’t tell you is that they don’t acquire much, if any, of the data themselves. Sure a few of us may connect directly to the Twitter API for basic Twitter data, but more often even for Twitter and certainly for blogs firms just buy it from an aggregator, mainly Gnip.
It’s often the first thing I tell a potential client who is using some other social media monitoring platform; we can use exactly the same raw data they already use and provide more useful information for those who want to go beyond Public Relations engagement.
Chris (Gnip): Gnip! We are providing social data to 8 of the 9 largest social media monitoring (SMM) firms. Also, some SMM firms connect with the public APIs offered directly from the publishers. The challenge is that public APIs weren’t typically designed to serve the challenges that SMM firms are trying to address. Instead, the public APIs were designed for more consumer use cases and thus have some limitations that can make social media monitoring use cases challenging to solve.
I often give the analogy that the public APIs were designed for fly fisherman and SMM firms are oceanographers. SMM firms need full coverage of real-time social data with ultra-low latency from sources that are reliable and will scale as their demands grow. The public APIs don’t typically fit in that mold.
Kami (Zoetica Media): Many of the vendors are getting their public data from the same sources, open APIs and through services like GNIP that provide a pipeline of data for both brands and social media monitoring group. Others, like Radian6 have additional agreements directly with social networking services to get the whole pipeline of data. Still others have their own proprietary crawlers to gather data.
The differences come in what they DO with the data. How do they filter out the considerable noise? How do they help the end user make sense of the data? How can the expense be justified? I think that there needs to be a much better rubric for comparing the plethora of services and match what they offer to the needs of each business
Nathan (Social Target): I haven't kept a count on how many companies use which methods, but everyone has two basic options: collect their own data from the original sources, or buy it from an aggregator. Collecting their own data creates an opportunity to compete on the quality of that collection, but it's a lot of work. Buying from an aggregator off-loads the work, at the expense of having a commoditized data stream.
I think most of the high-end vendors do some of both, buying from multiple vendors (for different types of data), while also spidering and pulling directly from feeds and APIs to fill out their coverage of targeted sites.
4. What is your opinion on the analytical tools that are currently available for social media analytics?
Kami (Zoetica Media): There is no one "killer" tool, but Google Analytics has turned on the heat with its free Google Analytics offering. Right now the tools seem bifurcated between data and analytics. I would love to see more vendors combine these two disciplines into one common dashboard.
Overall, I think that analytics are overwhelming. You can measure almost anything, but what matters? The current tools don't help with answering that one question and no one tool ever will define that. There needs to be a much deeper education on this topic to build an understanding about this with the end users of analytic tools.
Nathan (Social Target): I wish I had time to play with them all. :-)
I see a lot of very similar tools, especially the ones that are interesting primarily for the nationality of the vendor. I think it's a sign of the maturing of the market, that more countries can support a domestic vendor that offers industry-standard capabilities. The high-end tools have had very interesting capabilities for several years, but they require users to think about what they're looking at. If customers come to some of these platforms expecting easy answers on the first screen, I could see how they would be disappointed.
I'm interested to see companies that make it easier for less analytically-minded users to get to useful insights, but I'm more interested in the companies that are finding different ways of looking at the data. The whole area of influence and influencer analysis is very active lately, and I'm starting to see more geospatial analysis of social media data. Life's more interesting when I'm learning something new, and this market has been very good on that point.
Chris (Gnip): We’re still in the early days of the industry, but things are evolving quickly. I’m most excited about the diversity of solutions that are coming to market. We have hundreds of companies addressing different challenges and they are taking lots of different approaches. All the innovation that is happening in the SMM industry now will ultimately lead to some big wins for the brands that are using SMM insights to drive business decisions.
Tom (OdinText): I’m going to assume you’re asking me about real text and social analytics tools that you actually have to pay for, and not the free toys so often mentioned by ‘social media experts.'
Anderson Analytics started out by offering text analytics and data mining consulting, and therefore we licensed or partnered with some of the major text analytics software firms of the time. We were in effect put between these software firms and the fortune 500 clients who were asking “so what?!”. Having used so many tools, and developing 7 years of use case experience before developing OdinText was definitely helpful for us. There are a lot of good tools out there depending on the specific data and use case. We’ve worked hard to make ours both powerful and easy to use for our specific customers.
If you missed the previous posts in the interview series you can access them here: Part 1 Part 2
We look forward to seeing you at the Summit!