Continuous Intelligence: the Nexus of Data Integration, Analytics and Operations
To understand the power of intelligence, consider the Battle of Midway in June 1942. US Navy cryptographers correctly predicted Japanese fleet movements, helping dive bombers find and destroy three of their carriers. Five minutes of combat arguably changed the course of World War II.
How did they achieve this? Military personnel collected, transformed and analyzed a wide range of data points in context – and put those findings into action.
The emerging discipline of Continuous Intelligence seeks to achieve similar, if less dramatic, results in today’s business world by automating most of that work. While enterprises have a long way to go, machine learning, elastic cloud computing and oceans of data make this old idea more achievable than ever before.
Continuous Intelligence (CI) integrates historical and real-time analytics to operate, monitor and tune systems of all types - cellular networks, theme parks, factories, banking infrastructure, you name it. CI generally encompasses data ingestion, transformation, analytics and decision recommendations. It casts a wider net than traditional BI because it incorporates contextual data to understand factors such as market behavior, social media trends and economic indicators. It also shortens time to action by linking directly to operational processes, which in turn generate more data to analyze.
Done right, CI creates a virtuous cycle.
CI dovetails nicely with popular enterprise initiatives. It includes real-time data streaming and layers onto the cloud platforms to which enterprises are migrating. CI also applies machine learning and other advanced algorithms to engage customers and streamline operations in creative new ways. Perhaps most importantly, CI drives data democratization by putting actionable recommendations at the fingertips of analysts and operators throughout the enterprise.
The intended results: higher analytics value, faster response times, new sales and increased efficiency. Vendors such as Alteryx, Qlik, Splunk, Sumo Logic, Swim.ai and Tibco address this opportunity, each with a different architecture and technology mix. Common applications of CI include optimizing network performance, improving the customer experience, and monitoring of distributed and connected infrastructure. CI also applies to IT operations, DevOps and security.
Let’s explore an example use case to understand how Continuous Intelligence works. Suppose a retailer needs to implement and tune a new Back to School pricing strategy in the highly competitive market for backpacks. Its financial analysts build location-specific store prices for key regions across the United States and Europe, based on historical prices and recent purchase behavior, as well as contextual data about local COVID-19 infection rates and school plans for virtual/in-person learning. They go live, then monitor real-time purchase trends, both in-store and online, along with social media trends related to school planning, backpack shopping and COVID-19. They use a machine learning model to recommend ongoing, location-specific price adjustments based on all these data points as they are continuously ingested, transformed and analyzed.
Today CI use cases like these are more technology vision than business reality. Enterprises still struggle with high latency, manual scripting, siloed tools and disjointed workflows. Data analytics teams often use “CI” solutions for elementary use cases, such as automated data prep and reporting, refinement of marketing outreach lists and responses, or analysis of IT operational logs.
This immaturity reflects enterprise problems more than anything else. To make CI a reality, enterprises will need to address the challenges of heterogeneity, customization and data literacy.
Heterogeneity. IT and business teams often still struggle to integrate basic operational processes and achieve even rudimentary analytics insights across their diverse environments. Their applications, data platforms, data management tools and BI tools, accumulated through one-off decisions over decades, often need specialized coding to interoperate. New CI tools help patch things together with open APIs, but cannot fully correct for the proprietary limitations of legacy systems.
Customization. Enterprises often can standardize data ingestion, transformation and analysis, but still need to customize when tying back to operations. This is tricky because IP and operations vary widely by industry and even company. An offshore oil drilling platform requires different IoT systems than an automobile factory. A commercial bank requires different mobile applications than a retailer. Many of them lack the specialized programming skills this requires, and fail to build the right bridges between IT and operations technicians in the business units.
Data literacy. Many business managers and operational managers still lack the necessary skills to make more data-driven decisions. They need training on basic analytical concepts and how to map those back to their roles. They also need training on how to use CI and BI tools. Enterprises often neglect to invest in the right educational programs to make this happen.
Continuous Intelligence lays down a gauntlet. Data analytics leaders that invest in new CI tools – along with the necessary integration, customization and training – can gain competitive advantage in the turbulent COVID-19 era. Our next blogs will explore architectural approaches to CI, and how to navigate the trade offs it introduces to your organization.