Abundance of data is contributing to increasing pressures on how to quickly sift through vast amounts of information and hone in on the valuable insights. Add to that the fact that most organizations assume that the more data they have; the more insights are being overlooked; and the data-to- insights journey suddenly seems very overwhelming. The folks who walk around with T-shirts that claim “Torture the data till it speaks” don’t help matters, implying that it is only a matter of time till all data transforms itself to insights. So what is really needed to make this transformation and how can organizations quickly embark on a journey and arm themselves with the right information? So does all data have value? What value does it have? How do organizations go about intelligently harnessing the value from all “relevant data”…and how do you determine relevance?

1. Understand Business Objectives – When you look at how much data is available, determining what is needed by an organization and mapping it to business objectives becomes the most important part of determining the relevance of the data. At a later date, this very same mapping to broader organizational strategic initiatives helps get the buy-in for analytics solution deployment. Unless we understand what the business is seeking to do with the data, we cannot assess its potential. E.g. Businesses may look to increase sales/ make product recommendations based on POS data, reduce churn, or generate alerts and flags on some operational data. Once we understand the business objectives, we can do a preliminary level of exploratory data analysis to see if the data has potential to support such objectives (or a chosen modeling methodology to support them).

2. Structured Roadmap – Dealing with vast amounts of data from ever increasing sources and systems (internal and external) always leads to the question of where to start. Once the business objectives are understood, it is important to map them to the insights that can be reaped from the data to form a structured roadmap. Below is one such high level roadmap for retail, designed in conjunction with our Retail expert Pawan Joshi that can help organizations embark on a structured data to insights journey

3. Data Visualization and Exploration – Once we decide what to do with the data, data visualization tools can help us quickly understand it. Many of the data visualization tools today offer such rich outputs that they truly have become the first window to insights. Additionally, many of these tools have significantly reduced data exploration times, decreasing modeling life cycles and making it easier for a data scientist to get insights faster.

4. Deployment strategy – Insights are useless unless they are tied to actions. Delivering insights is the final outcome in the complex journey of data to insights, and demands integration and technology deployment. Increasingly, analytics are just a small part of the elaborate technology delivery architecture. For example, if a retailer makes recommendations to customers as they are shopping online or in a smart cart in a store, then we need access to the most current data, a robust recommendation engine and all associated technologies.

The data to insight journey is constantly evolving with newer data sources becoming available all the time. Business innovations, sometimes fuelled by analytics, may again lead to additional data that could be used to gain further insights leading to a continuous cycle of data to insights. For example, recommendation engines are propelled by analytics. Once deployed, a recommendation engine can help gain further insights to customer behavior and evolving trends. One of the latest entrants in retail innovation is the use of robots. Imagine the potential of collecting customer, store and product data from robots continuously parading the store aisles…and it will not be long before somebody asks, when could we start seeing the insights from the data that has been collected. So while we certainly don’t prescribe torturing the data, we can keep looking for it or collecting additional data until it speaks to us.