Businesses understand the urgency of communicating with customers in language that resonates. As part one of this SearchCIO Trailblazer profile showed, the rise of Twitter as a powerful communications channel means that businesses increasingly have to learn to talk tweet to keep up with customers. Analyzing tweets for potential business value is no easy task, however, requiring both qualitative and quantitative expertise. One of the biggest social analytics challenges businesses will face is staying on top of language changes to ensure that they don't miss part of the conversation.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Just ask Sherry Emery, senior research scientist at the University of Illinois at Chicago's Institute for Health Research and Policy. For Emery, who studies how smoking habits and products evolve, keeping up with how people talk about brands can mean the difference between meaningful and bogus results. An example is the word square, a slang term for cigarette that initially escaped her notice as a key word to search. "Key words are the fishing hooks that you get all of the data with," she said. "So, you have to choose the correct key words in order to get the information you're after. If you don't, then you're missing important pieces of the conversation."
Emery, for one, considers herself lucky to be working with college-age research assistants who help her stay on top of emerging terms, but her team also includes an eclectic mix of experts. Economists; biostatisticians; and people trained in computer science, medical anthropology, social work and even communications round out the group.
"What we're talking about, really, is content analysis or, traditionally, the qualitative research done by anthropologists and sociologists," Emery said. "But now we're doing it at a quantitative scale." Those who can speak both languages -- and act as interpreters between the two worlds -- are essential to the ultimate meaningfulness of the research study.
But even with all that expertise on board, a term as simple as smoking can be problematic for researchers, she said. Smoking a cigarette, for example, is very different from smoking ribs. So, one of the challenges Emery and her team encounter is figuring out how to capture the data they want, clean it and make sure it's reliable enough to analyze. Researchers and businesses aiming to derive value from tweets also have to consider the context of the tweets, including how the tweets might relate to an event playing out on a larger stage, stressed Carol Haney, senior vice president for group product marketing at Toluna, a London-based survey technology provider.
When a colleague of Haney's decided to map out the Twitter mentions of Salvia divinorum, a plant that is the source of a psychoactive drug, most of the results were unsurprising and followed trends he was already tracking in his research. That is, until 2010 when mentions of the drug on Twitter and other social media sites went through the roof. "The spike is when Miley Cyrus was videotaped actually taking the drug. People went crazy on Twitter for a bit," Haney said.
IT involvement critical to the success of tweet analysis
Social analytics tools
Gartner Inc.'s 2013 "cool vendors" in social and content analytics included just five. Here's how the consultancy described each vendor on the list:
1. ATIGEO: "An information analysis platform that discovers relationships, patterns and connections between different information objects in large volumes of data where it has been traditionally difficult to identify linked concepts."
2. METAVANA: "A sentiment-analysis platform, which collects and processes social and Web data aimed at specific industries."
3. NETBASE: "An enterprise social intelligence platform that monitors what people are saying on social media. It scans a wide variety of social media sites (such as Facebook, LinkedIn, Twitter and Pinterest); blogs; forums, and news sources. The information is stored in an accessible social intelligence warehouse."
4. RAMP: "Focuses on talk tracks and video segmentation (not object recognition), which gives media companies better facility to make their broadcast (or narrowcast) programming more searchable and more valuable to advertisers. It develops metadata synchronized to video timelines and content, which allows users to find what they want without the collaborative filtering. Now Ramp looks to develop an on-demand enterprise video platform to augment Microsoft SharePoint."
5. YARCDATA: "Uncovers previously unknown patterns and relationships across large repositories of multistructured data and represents one of the biggest opportunities to derive value from analytics. YarcData enables a flexible way to model and update a schema on-the-fly to conduct real-time, iterative pattern discovery; what-if analysis; and inferencing and deduction on large (up to 512 TB in memory so far) and dynamic data represented in unpartitionable graphs."
Source: Gartner's Cool Vendors in Content and Social Analytics, 2013
For businesses, understanding the broader context of social information also means looking inward at their own processes. Companies that analyze social analytics still tend to work in silos; for a social analytics program to really flourish, social data needs to be integrated with a company's more traditional data sources, according to Jenny Sussin, an analyst at Stamford, Conn.-based consultancy Gartner Inc.
Doing so paves the way for things like multi-channel analysis or integration with business intelligence (BI) applications, which can provide richer, deeper insights into trends and customer behavior. By integrating social data into legacy customer data, IT can add significant value to marketing, sales, customer service and any other department relying on social analytics today. Multi-channel analysis, for example, achieves the greatest levels of success when IT is involved, Sussin said.
The University of Illinois' Emery, for example, relies on an arsenal of service providers to gather and manage the data. She uses Gnip, a data service provider, to pull the raw data from the Twitter application programming interface. The data is fed through DiscoverText, a Software-as-a-Service provider that enables Emery and her team to classify and clean the data.
In many corporations, lines of business use similar tools to parse through and explore social data, often making those investments without consulting IT, according to Sussin. That doesn't mean the social analytics door is closed to CIOs and IT leaders, she added.
One of the things IT can do is create guidelines for how social analytics data is kept and how it should be integrated with customer records. "All of that information around data, data storage, data protection, that's where the real opportunities for IT to get involved are," Sussin said. But taking on the role of data governance is not the only way IT can add value to a social analytics program. She also suggested that IT offer to participate in contract negotiations with SaaS providers. For some CIOs, that may require adjusting their technology strategy because, unlike purchasing a new database that will be around for the long haul, social analytics tools are young and finding the right fit isn't always easy.
"If you try to identify a solution as a life-long solution for a social media application, you're not going to find it," Sussin said. "A lot of these solutions are new to marketing, and most of these solutions are still working out their integration strategies."
A CIO or IT presence at the negotiating table to ensure that investments are compatible with legacy applications or if appropriate, to sign off on what everyone agrees is a disposable solution will help the business in more ways than one, Sussin said. "When IT does get involved, the services cost associated with … social application proposals goes down," she said.