Reading articles about data anonymization makes it clear that it is not an entirely effective security measure (here and here), but still part of a robust security capability, and required if your organization is affected by GDPR. (I use “anonymization” as a general term encompassing techniques that de-identify personal data within a given data set.)
But there’s a positive side of anonymized data that hasn’t received much press. Providing anonymous data to senior managers who don’t need access to personal data can encourage them to take a broader perspective, and thereby bring new energy to fact-based senior planning and analysis.
Reading available success stories, analytics seems a business of detailed rather than big picture wins. For example, the personalization movement of the 00’s focused analysis to the individual level, and all of CIO Magazine’s recent top ten list of analytics successes that are specific enough to count involve operational improvements. Many prominent analytics wins are “inside the box” successes that reduce costs or improve current processes rather than insights that revolutionize entire ways of doing business.*
There’s strong gravity-to-detail in many of the successful analytics ventures I’ve been involved with. Availability of precise, timely operational data vastly improves ability to meet and exceed goals, so detail-focused staffers drive requests for visibility to the details of processes they manage. Therefore, requests for analytical services focus on predicting an individual’s buying patterns, increasing a given physician’s outreach to patients with chronic illnesses, speeding up service deliveries, and so on. The volume and urgency of operationally focused requests can displace efforts to serve higher level needs.
On the other hand, there are many cases of transformative use of analytics. “Digital native” companies are the best examples: Capital One transformed the credit card industry by recognizing there’s money to be made by lending to higher risk customers; Amazon revolutionized retail with data-first approaches to all aspects of its business, and so on. But there are also success stories from legacy organizations. Lockheed Martin harnessed “dark data — corporate data that could be useful to a company but is, instead, gathering dust in storage — for more proactive project management,” improving program foresight appreciably. Nissan aggregated responses to a customer completed “request form” across their many localized websites to “paint a vivid picture of vehicles’ popularity in a particular region [enabling] advertising campaigns and production to be tailored to the needs of a region instead of a country or continent as a whole.”
The Nissan and Lockheed Martin examples show how a big picture view can enable benefits beyond current process improvements. However, enabling a broader perspective is not just a matter of cutting off senior access to detailed data. Here are a few recommendations:
- Build a parallel anonymized, detailed database: The shift in perspective isn’t from detail to summary, but rather from focusing on the individual to focusing on relationships among individuals. So the question changes from “why did this person buy a bluetooth speaker”, for example, to “why is purchase of fitness equipment correlated with purchase of bluetooth speakers?” The point is to provide complete detail in order to support any possible slicing, dicing, and analysis.**
- Provide big picture analytics examples: Seed senior managers’ imaginations with a few “killer viz” examples. Established analytics teams will likely already provide high-level KPIs, if not these should be developed. But in addition, it’s important to find one or two insightful visualizations to spur imaginations, in the same way that early big data efforts built the killer proof of concept app.
- Soften the blow during transition: Senior managers might see migration to anonymous data as a loss of privilege, which no one enjoys. As always during such transitions, communication is key. Tell them about it early and often, emphasizing the power of the anonymous perspective and reduced threat of loss of personal data. Be prepared to make exceptions for those who in fact do need full data access, or even holdouts who don’t support the transition. Make available accessible and relevant orientation and training on the scientific method, generating hypotheses, and basic statistics for those so inclined.
If you make that transition carefully, with senior managers involved along the way, you stand a chance of sparking the fundamental improvements promised by big data-enabled analytics.
*One way to tell if your analytics team primarily serves operational needs is to look at your visualizations: if they tend to be grids with numbers rather than charts and graphs, then that’s probably the case.
**Building an anonymized database is tricky, involving a number of database design considerations. For example, if your relational joins are based on personal identifiers, the anonymized version will either have to use different columns, possibly added surrogate keys, or replace those key values consistently across the database. That said, the effort is well worth it if anonymization spurs fact-based senior leadership, not to mention the incremental reduction in exposure to loss of personal data.