The analytics market has expanded. Where there was once only reactive analysis, prescriptive options have arrived, which are able to identify midsize issues and suggest timely correction. Predictive analytics has seen an upswing in adoption thanks to lower initial costs and higher return on investment (ROI); but getting the biggest dollar value back means dealing with unique predictive concerns, most importantly data privacy.
Privacy by Design
Privacy by design is the ultimate goal for predictive analytics solutions, according to a recent article at Data Informed. Instead of adding in privacy tools after a business intelligence (BI) deployment, an entire industry has sprung up to integrate data privacy into predictive software before it reaches company computers. Jeff Jonas of IBM talks about his efforts in this field, which include the development of de-identifying technologies. The idea behind de-identification is to obscure pertinent information about an individual before using the person’s data as part of an analysis, and many solutions compel midsize companies to first “match” their data and to strip customer records of personally identifiable information (PII). Jonas’ tool instead modifies these values without removing the identity of a record entirely; data such as name and address are retained while more sensitive information is modified to be nonhuman readable and nonreversible. Here, the integrity of data sets remains intact, allowing full use of predictive analytics, but data privacy is maintained. For consumers already ambivalent about the idea of their information being used to predict future behavior, this kind of obfuscation is quickly becoming a necessity.
Recent data from an Information Management survey shows that predictive analytics solutions are on the rise. Sixty percent of respondents said that they already used at least one predictive analytics tool in the cloud, while a whopping 90 percent were interested in deploying such a solution within the next few years. Most telling, however, was that two-thirds of those asked reported a “positive impact” from their analytics tools in 2013, in large measure thanks to lowered cost. Market demand has forced down the purchase price of BI solutions while keeping ROI steady, producing a net increase. According to the study, the most important area of focus in these deployments is customer engagement, including satisfaction, profitability and retention.
For midsize IT professionals, this dovetails perfectly with the notion of data privacy. Big data-driven, predictive tools are being used to increase consumer satisfaction and engagement; but without the right kind of privacy protection, these solutions can do more harm than good. Used as a way to anticipate shifts in the larger market or to help a company properly position its brand, such tools are benign; used to (seemingly) target specific individuals, they come dangerously close to stepping on privacy rights. Ultimately, midsize admins should be prepared for increased executive interest in predictive technologies, and they should choose wisely when the time comes for deployment: Which solutions include baked-in identification protection rather than an expectation of third-party privacy? In addition, companies should never rely on the strength of a product’s claim alone; internal expectations for privacy must be clear and concise, and if a BI tool cannot deliver, it is better to find an alternative than to risk ROI at the hands of ineffective data privacy.
By: Doug Bonderud
Originally published at midsizeinsider.com
A freelance writer since 2009, I have a particular passion for technology and its impact on our daily lives. As an evolving resource, technology changes us as much as we inform its development, providing fertile ground for thought.