Print advertising is down, digital advertising is up and social media is emerging as the hottest hot spot in online marketing. But how can you understand what's being said by customers across social media, and then respond either individually or in broad public relations or advertising campaigns?
It's a challenge specialized firms and analytics leaders like IBM, SAP and SAS are all trying to solve. And don't forget Salesforce.com, which recently announced it
The stakes in this market are growing. U.S. advertisers will spend more than $2 billion on social media sites this year, up 24% from 2010, according to research firm eMarketer. It's a small but fast-growing slice of total spending, compared with magazine and newspaper spending in the U.S. pegged at $38 billion and all forms of online advertising totaling $28 billion last year, according to Winterberry Group figures.
These days I can't seem to stay in a hotel or rent a car without getting a follow-up survey by e-mail. Sentiment analysis technologies are undoubtedly at work behind the scenes compiling responses to structured questions -- "How clean was your room on a scale of one to ten?" -- and correlating the results with open-ended, textual-response requests like "Please describe why you were dissatisfied with the cleanliness of your room."
But companies know it's much more important what you tell your friends in Facebook than what you tell the company via a survey. That's why interest in social-media analysis is growing and has led to a spate of new products.
IBM Cognos Consumer Insight (CCI), an on-premises software product introduced last week, is the latest example, and it's one of several products and services IBM has for analyzing human language. CCI combines multiple IBM technologies. IBM's Hadoop-based Big Insights engine provides database-like management capabilities geared to text. That's needed because social network comments can't be neatly ordered in columns and rows like structured data. Language-analysis technologies from IBM's Almaden Research Lab flag topics, and positive and negative sentiments in comments. The application is also integrated with predictive-analytics technology from SPSS, the analytics firm IBM acquired in 2009.
The product is expressly designed to help marketers monitor social media to understand what consumers are talking about and what they consider most relevant to brands and products. The connection with predictive analytics "helps you understand what types of campaigns are likely to give you the biggest movement in sentiment," says Deepak Advani, IBM's vice president of predictive analytics. The idea is to understand what topics matter most to important customer segments, such as loyal customers or most-profitable customers, so you can then craft advertising, public relations and social media campaigns that will have the biggest impact on the bottom line.
That's the theory, at least, but IBM could not cite any beta customers that have proven the approach.
The use of technology to sift through vast quantities of text-based content to understand customer sentiments is not new. The focus on social media has grown with the rise of social networks like Facebook, Twitter, and LinkedIn. Specialized companies including Converseon, Cymfony, Lithium, Nielsen Buzzmetrics, Radian6, and Visible Technologies have been among the pioneers, according to Seth Grimes, an analytics expert who organized the recent Sentiment Analysis Symposium.