The Many Dimensions of Social Media Data
By Tim Wilson on in Analysis, Data Management, Social Media with No Comments
I’ve been thinking a bit of late about the different aspects of social media data. This was triggered by a few different things:
- Paul Phillips of Causata spoke at eMetrics in San Francisco, and his talk was about leveraging data from customer touchpoints across multiple channels to provide better customer relationship management
- I’ve been re-reading John Lovett’s Social Media Metrics Secrets book as part of an internal book group at Resource Interactive
- We’ve had clients approaching us with some new and unique questions related to their social media efforts
What’s become clear is that “social media analytics” is a broad and deep topic, and discussions quickly run amok when there isn’t some clarity as to which aspect of social media analytics is being explored.
As I see it, there are four broad buckets of ways that social media data can be put to use by companies:
No company that is remotely serious about social media in 2012 can afford to ignore the top two boxes. The bottom two are much more complex and, therefore, require a substantial investment, both in people and technology.
Now, I could stop here and actually have a succinct post. But, why break a near-perfect (or consistently imperfect) streak? Let’s take a slightly deeper look at each bucket.
(I almost labeled this bucket “Community Management,” but the variety of viewpoints in the industry on the scope of that role convinced me to leave that can of worms happily sealed for the purposes of this post.)
Social media requires a much more constant intake and rapid response/action based on data than web sites typically do. Having the appropriate tools, processes, and people in place to respond to conversations with appropriate (minimal) latency is key.
Key challenges to effectively managing this aspect of social media data include: determining a reasonable scope, being realistic about the available on-going people who will manage the process, and, to a lesser extent, selecting the appropriate set of tools. Tool selection is challenging because this is the area where the majority of social media platforms are choosing to play — from online listening platforms like Radian6, Sysomos, Alterian, and Syncapse; to “social relationship management” platforms like Vitrue, Buddy Media, Wildfire, (Adobe) Context Optional, and Shoutlet; and even to the low-cost platforms such as Hootsuite and TweetDeck. These platforms have a range of capabilities, and their pricing models vary dramatically.
Ahhh, performance measurement. When it comes to social media, it definitely falls in the “simple, but not easy” bucket. And, it’s an area where marketers are perpetually dissatisfied when they discover that there is no “value of a fan” formula, nor is there “the ROI of a tweet.” But, any marketer who has the patience to step back and consider where social media plays in his/her business can absolutely do effect performance measurement and report on meaningful business results!
Chapters 4 and 5 of John Lovett’s book, Social Media Metrics Secrets, get to the heart of social media performance measurement by laying out possible social media objectives and appropriate KPIs therein. High on my list is to make it through Olivier Blanchard’s Social Media ROI: Managing and Measuring Social Media Efforts in Your Organization, as I’m confident that his book is equally full of usable gems when it comes to quantifying the business value delivered from social media initiatives.
When it comes to technologies for social media performance measurement, we generally find ourselves stuck trying to make use of the Operational Execution platforms. They all tout their “powerful analytics,” but their product roadmaps have typically been driven more by “listenting” and “publishing” features than they have been driven by “metrics” capabilities. With Google’s recent announcement ofGoogle Analytics Social Reports, and with Adobe’s recent announcement of Adobe Social, this may be starting to change.
Leveraging social media data to improve customer relationship management is something that there has been lots of talk about…but that very few companies have successfully implemented. At its most intriguing, this means companies identifying — through explicit user permission or through mining the social web — which Twitter users, Facebook fans, Pinterest users, Google+ users, and so on can be linked to their internal systems. Then, by listening to the public conversations of those users and combining that information with internally-captured transactional data (online purchases, in-store purchases, loyalty program membership, email clickthroughs, etc.), getting a much more comprehensive view of their customers and prospects. That “more comprehensive view,” in theory, can be used to build much more robust predictive models that can let the brand know how, when, and with what content to engage individual customers to maximize the value of that relationship for the brand.
The challenges are twofold:
- Consumer privacy concerns — even if a brand doesn’t do anything illegal, consumers and the press have a tendency to get alarmed when they realize how non-anonymous their relationship with the brand is (as Target learned…and they weren’t even using social media data!)
- Complexity and cost — there is a grave tendency for marketers to confuse “freely available data” with “data that costs very little to gather and put to good use.” Companies’ customer data is data they have collected through controllable interactions with consumers — through a form they filled out on the web, through a credit card being run as part of a purchase, through a call into the service center, etc. Data that is pulled from social media platforms is at the whim of the platforms and the whim of the consumer who set up the account. No company (except Twitter) can go out to a Twitter account and, in an automated fashion, bring back the user’s email address, real name, gender, or even country of residence. It takes much more sophisticated data crawling, combined with probabilistic matching engines, to get this data.
Despite these challenges, this is an exciting opportunity for brands. And, the technology platforms are starting to emerge, with the three that spring the most quickly to my mind being Causata, iJento, and Quantivo.
Trend / Opportunity Prediction
This is another area that is really tough to pull off, but it’s an area that, admittedly, has great potential. It’s a “Big Data” play if ever there was one — along the lines of how the Department of Homeland Security supposedly harnesses the data in millions of communications streams to identify terrorist hot spots. It’s sifting through a haystack and not knowing whether your’re looking for a needle, a twig, a small piece of wire, or a paperclip, but knowing that, if you find any of them, you’ll be able to put it to good use.
The wistfully optimistic marketing strategist describes this area something like this: “I want to pick up on patterns and trends in the psychographic and attitudinal profile of my target consumers that emerge in a way that I can reasonably shift my activities. I want an ‘alert’ that tells me, ‘There’s something of interest here!'”
It’s a damn vague dream…but that doesn’t mean it’s unrealistic. It’s a multi-faceted challenge, though, because it requires the convergence of some rather sticky wickets:
- Identifying conversations that are occurring amongst people who meet the profile of a brand’s target consumers (demographic, psychographic, or otherwise) — yet, social media profiles don’t come with a publicly available list of the user’s attitudes, beliefs, purchasing behavior, age, family income, educational level, etc.
- Identifying topics within those conversations that might be relevant for the brand — we’re talking well beyond “they’re talking about what the brand sells” and are looking for content with a much, much fuzzier topical definition
- Identifying a change in these topics — generally, what marketers want most is to pick up on an emerging trend rather than simply a long-held truism
To pull this off will require a significant investment in technology and infrastructure, a significant investment in a team of people with specialized skills, and a significant amount of patience. I chuckle every time I hear an anecdote about how a brand managed to pick up on some unexpected opportunity in real time and then quickly respond…without a recognition that the brand was spending an awful lot of time listening in real-time and picking up nothing of note!
This area, I think, is what a lot of the current buzz around Big Data is focused on. I’m hoping there are enough companies investing in trying to pull it off that we get there in the next few years, because it will be pretty damn cool. Maybe IBM can set Watson up with a Digital Marketing Optimization Suite login and see what he can do!