Data is meaningless unless it helps make decisions that have measurable impact. Generating value is a matter of connecting data to insights to action in a fast, repeatable way.
Data is meaningless unless it helps make decisions that have measurable impact. Unfortunately, many decision makers are ensnared rather than enlightened by Big Data, preventing data and insights from making it to the front lines in relevant and usable forms. Too many Big Data projects are formulated without input from front-line operators, or consume so much time that the insight goes stale before it can be used.
In our experience, generating value from Big Data is a matter of connecting data to insights to action in a fast, repeatable way. Picture a factory.Insights are products—goods that are valuable because they are useful; data is the raw material from which the products, the insights, are made; and front-line operators are the consumers, or the people who need and use the product.
The “insight factory” approach enables companies to sift through massive amounts of data quickly, run the right analytics, and provide relevant insights so people can take meaningful action. And we’ve seen top-line sales increase 5–15 percent as a result.
Here are the four steps to get there:
1. Decide what to produce
Before work begins at an insight factory, you should have a clear understanding of what you want to achieve, such as reducing customer churn or predicting what a given customer segment will buy next. Decide what discrete questions your business needs to answer and the actions you want those answers to enable. Prioritize questions that address the largest economic opportunities and that lead to practical actions. Then configure your factory to produce just those insights. One retailer, for example, discovered that 90 percent of its year over year sales decline was concentrated in 12 percent of its customers in specific markets. It focused questions, accordingly, on understanding the root cause and quickly reversed the trend with targeted local market merchandising tactics.
2. Source the raw materials
While it’s useful to identify a range of data sources to build insights, start with the best data immediately available. Chasing after the “perfect dataset” is time-consuming (and often fruitless) and reduces the ability to act quickly. Instead, start with “small data”. A comprehensive “data warehouse” is a great asset over the long term, but a smaller, more selective “data mart” makes it easier to produce insights fast, preventing you from getting mired in complexity. Over time, you can then layer on additional data sets. In one case, a leading retailer setting out to understand its customers began by complementing transactional POS data with third-party customer data from aggregators, syndicated competitor data, and public sources that were immediately available. Over a year, it enriched these insights by adding social media data (for sentiment analysis), location data (to understand store traffic and movement), and financial information from credit card providers (for share-of-wallet).
3. Produce insights with speed
We have found that when it comes to analytics, productive action is mainly a product of speed. Focusing on quick decisions and execution, which circumvent long discussions, leads to insights the front line can actually use. Put finite time limits on your insight factory to force short production times and rapid bursts of structured output based on repeatable analytical models and automation.
We recommend acting like a start-up. Start-ups are driven by an inherent need for speed that doesn’t let perfect get in the way of good enough. Aware that a futile quest for perfection creates paralysis, they thrive on a test-and-learn culture that celebrates failing early and moving to action quickly with imperfect information. Create small, nimble teams combining strategic, analytical, and technical skills to address specific topic areas rather than a single, generalized, and usually slow-moving “committees.” To keep the factory running around the clock, consider recruiting offshore talent to execute structured analysis continuously, at relatively low cost.
4. Deliver the goods and act
“Good enough” information available now can be used now to inform specific actions. If data yields the insight that milk and eggs are 90 percent likely to be purchased together, why not quickly pilot the placement of milk and egg shelves next to each other rather than wait for more comprehensive options?
Making sure that insights drive action requires a clear understanding of what front-line managers can actually use. These managers need to identify what they need. Too often, marketers or sales people are provided with data analysis they subsequently ignore. In many cases, the analysis isn’t practical, isn’t clear, isn’t trusted, or isn’t perceived as relevant. For an insight factory to work, think of the sales and marketing people who use the insights as your customers. They need to be part of a process that gives them simple ways to use the insights, such as interactive frontline tools (e.g. competitive price tracker, customer scorecards, or store operations health monitor). The most effective approach is not to push these tools on managers, but to listen and respond to their needs and then create pull.
Build a “factory” culture over time
To successfully weave the insight factory into the fabric of the way the business works, avoid leading off with momentous change. Accustom stakeholders to incrementally embed data and insights into everyday decision making. Over time, the integration of insight factory production into business-as-usual will create a willingness to accept bigger decisions and greater change.
This article originally appeared on the Forbes website