Analytical hacking turns an instinctive approach based on hunches into a numbers-driven game—in four key steps.
First, get visibility of the meaningful data sources; for example, product sales performance, calendar and email data regarding who is communicating with whom and when, sentiment analysis regarding team morale and stress levels, development costs, and repeat conversions.
Second—define the question you want to ask the data; for example, what sets apart the sales teams and products which have the highest success rates?
Then run a split test or experiment, comparing your different options; Product A’s development reviews were held on a Tuesday morning, while Product B was Friday afternoon. What happens to product development when review meetings are no longer held at a time when minds wander to the weekend? Other experiments could include: What happens when people meet in person as compared to remotely in a conference call? What happens when meetings are kept to a small number of participants as compared to a larger number?
Depending on the results, you can scale and implement the successful experiments "into the wild."
However, it doesn’t end there. This process is repeatable, and can become a continuous, closed-loop approach to proactively challenge and improve organizational performance. (Think ZBB, but this time it’s ZBO: Zero Based Organization.)
What part of the company is going to run this ongoing capability?
A strong contender would be a multi-disciplinary “HR Organizational Effectiveness” team that includes a combination of HR, behavioral science, analytics and other expertise to provide the necessary combination of capabilities. This team would operate with the following principles.
Break out of your ERP Box
The sources of data available enable bigger questions to be asked, and the data experiments being used are different and more intriguing—with bigger impact: increased engagement and efficiency at an individual level, and better innovation, agility and productivity at an organizational level.
You can go well beyond traditional ERP systems and HR databases. Companies can now mine a range of unstructured data like e-mail, online activity, or smart sensors in wearables to paint a vivid picture of activities, emotions, relationships, performance, productivity and collaboration patterns that can be tied to newly available performance data at the individual, team, organization and enterprise level. (See sidebar about privacy considerations.) Organizations are collecting much of this data already; AI can now mine it for insights. Thirty-four percent of companies are using AI to find hidden value in so-called “dark data,” or digital information that is collected but not being used.1 By combining the power of three—data, analytics, and artificial intelligence—organizations are now able to realize new heights in value. Nearly four in ten companies (37 percent) now consider analytics to be an important capability to combine with AI.2
Choose your Goal
Create some hypotheses about causality and design high-quality experiments. Identify the business goals you are trying to achieve: Greater sales, more innovation, improved cost to serve, reduced time to market and so forth.
Let’s say you want to reduce time to market for product development. Your hypothesis here might be that stronger ties and better informal collaboration across departments such as research, development, manufacturing, marketing and sales would improve this metric. “Size the prize”: Challenge the business and show them the potential benefits.
Experiment and test. Repeat.
Determine what data you need to test your hypotheses. In this case, you could gauge the digital water-cooler through analysis of e-mail and social media sentiment. You could also compare different social network analyses of two separate product lines against each other.
One consumer packaged goods company shaved almost a year off its product time-to-market by identifying bottleneck employees (a few people responsible for connecting all others) and then redirecting communications to others with the potential to be good connectors, creating better collaboration and stronger ties among functional groups.
When actionable insights arise from the data experiment, apply a solution to one part of the organization. If successful, scale it to the rest of the company. Follow the seams of opportunity and bank the cash from results that point to better organizational health and performance.
Then run experiments for new sets of hypotheses. The experiments don’t have to be sequential, but can be run in parallel.
Sidebar: Hacking while respecting data privacy
The advent of laws like General Data Protection Regulation (GDPR) are shining a spotlight on the responsible use of all data.