As we enter 2015, big data remains a hot topic among marketers and customer intelligence professionals. While big data companies promised unprecedented rapid access to actionable insight, most big data projects actually fail. A mere 35 percent of companies with big data projects describe their initiatives as either “successful” or “very successful,” according to one report.
So why aren’t companies seeing ROI from big data? This week’s reads offer three potential reasons:
- Big data is not the right solution.
Contrary to what many companies say, big data projects require an investment in time to collect and analyze the data. That’s according to John Foreman, Data Scientist at the email software company MailChimp, who says it’s important for data scientists to recognize the limitations of big data.
“Most data science models that are of any degree of sophistication, require stability,” writes Foreman. Big data requires stable inputs, so the latest analytics tools can’t necessarily accelerate these projects.
He adds, “It's very hard for sophisticated analytics software and techniques running on ‘big data’ to run out in front of your changing business and radically benefit it.”
If you’d like to get insight that keeps up with the speed of business, big data may not be the answer. Foreman’s advice to companies is to get back to basics: thoughtful analysis of data “can identify what's happening in the business, and can communicate their analysis in that chaotic context,” he concludes.
Data scientist @john4man on why most #bigdata projects are not quick: http://ow.ly/J22Oq Part of @VisionCritical roundup. (CLICK TO TWEET)
- Your data is in silos.
Letting data collect dust won’t lead to insight. Various departments in your company need to collaborate on big data projects, determine what goals you collectively need to address, and then break out of data silos.
Talking to Dell Tech Page One, Vision Critical CMO Tyler Douglas says companies need to let various streams of data talk to each other—and customer intelligence needs to be part of the mix.
“CRM data is being gathered and analyzed separately from social media information, and customer intelligence data often sits in a different area of the business altogether,” says Douglas. “Seldom do these data streams work together.”
Letting data collect dust won’t lead to insight. Learn why #bigdata projects often fail: http://ow.ly/J22Oq (CLICK TO TWEET)
- It lacks the human touch.
Stephen Trask, a principal at emotion quantification company emoanalytics, says 92 percent of big data practitioners and experts do not use the word “emotion” when discussing experience and engagement metrics. Why is that a problem? Here’s Trask’s explanation:
With no “emotional compass”, the objective analysis in big data can be very accurate in defining a transaction or experience. But the question of, Will someone remember the experience? exists only in the emotion data. The emotion data is what prompts someone to blend the facts with perceptions and possible implications to the experience. That is how the black hole affects predictive modeling for recall.
For big data to deliver insight, companies need to blend emotional data with traditional data. Engaging with your customers directly—and consistently—is key to getting the necessary emotional data.
As these recent articles show, the first step toward seeing ROI from big data is recognizing its limitations, and then tactfully employing it within a broader context of specific business objectives. While big data is often touted as a mystical tool that magically gives companies insight into the motivations and attitudes of their customers, it can be, in fact, an incomplete and arduous tool that often produces more noise than signal. In the age of big data, using centralized and scalable platforms that directly engage with your customers has never been more critical— especially those which can support the analysis of disparate data streams while also allowing for direct engagement with the customer. Big data doesn’t eliminate the need to strengthen the customer relationship.