Leveraging analytics across the enterprise

Unlocking the value of data will be more and more a part of the digital transformation of enterprises.

abstract data analytics
Thinkstock

Digital transformation is driving new business models – stickier value propositions that drive top line growth, and large, streamlining opportunities that take out operating costs. In the emerging competition between the disruptors and disrupted in this new economy, unlocking the value of the data that sits across an existing book of business is becoming the new value driver for large enterprises. As a result, in 2018, we will see more businesses looking to cash in on this data currency.

Those that succeed are the companies that know how to prepare data for consumption, and use deep statistical inferencing and machine learning techniques to uncover smarter insights. And then, know how to transform these insights into action through active nudging – providing the right guidance at the right time, at the right stage in the process. Let’s explore how companies can make the most of their data, apply analytics and artificial intelligence (AI) to generate real, impactful insights, and then drive meaningful change in their organizations.

Start with a clear vision and goal

It is important to have the right start. An effective analytics strategy addresses more than data and insights – it includes people, process, and change management. Too often, we see fantastic insights end up underutilized by businesses that did not invest in a broader program to drive analytics into the core of their business. Industry evidence clearly shows a business-driven center of excellence around analytics always delivers more than a technology-driven analytics group.

Moving to the core of the analytics work, it starts with data strategy – design thinking through what data is needed, and making thoughtful choices that ensure processes and technology systematically capture the data. For instance, we see elevator companies now capturing load and stops at different times of the day so they can deliver better user experiences – data that we never captured before. Or, manufacturers capturing data from sensors on aircraft engines so they can optimize asset maintenance. Combining strategically important new data with other existing data from both internal and external sources creates a universe of rich, big data that businesses can draw upon.

Extracting, structuring and engineering data

Data engineering is the foundation of the analytics practice and involves data architecture (discovering, understanding, sourcing, and housing the data); data orchestration (ingestion, cleansing, transforming, and unification of data); and data governance (master data management, security, and provenance), making it consumption-ready for running modeling techniques to deliver business insight.

In addition, new advances in AI are opening up a treasure trove of information hidden in unstructured data. This is information housed in emails, PDFs, and other documents that Natural language processing (NLP) and computational linguistics algorithms can now convert into structured data for use. As a result, documents that previously had to be read by humans can now be automatically accessed, and data embedded in those documents can now be used for running analytics. For example, we see banks reading balance sheets using AI and deriving risk scores on lending portfolios dynamically. We see corporations reading procurement contracts dynamically and automatically reconciling them with invoices.

Generating insights from data

Data, once made consumption-ready, can then be put to work with various statistical inferencing techniques. The ever-expanding world of data science now provides some really mature techniques for optimization, classification, or clustering, which all help with prediction of outcomes. Optimization allows industrial manufacturing companies to predict the best placements of spare parts closest to high-end machine assets. Classification allows insurance companies to quickly look at incoming claims requests and route them dynamically for fast remediation. Clustering allows banks to look at fraud detection and anti-money laundering by detecting anomalies with higher accuracy.

Machine learning algorithms have progressed significantly in recent years, especially through the development of deep learning and reinforcement learning techniques based on neural networks. Computing capacity has also improved, so now it is possible to train larger and more complex models even faster. Finally, as the data we are capturing grows, our ability to train machine learning algorithms is improving, making it more effective for business use.

This fast-evolving world of machine learning is now entering the business world. For instance, we are finding machines that can accurately predict the cost of repairs for a damaged roof based on photographs, once they have been trained with millions of photographs and associated claims information. As the speed, accuracy, and cost to deliver insights becomes more favorable, data science and machine learning is opening up completely new business models.

Pair domain knowledge and process understanding with data science

One of the key components of making analytics actionable is to pair it with domain knowledge and process understanding. Contextualization and distillation are key in data sciences, and goal orientation is fundamental in machine learning. These come best from a deep understanding of the business context and the process handshakes happening upstream and downstream. These require the involvement of the right personnel who can add these critical components. Too often we have seen projects underperform because they lack this important foundation.

Turn analytics to action

As Lisa Morgan recently shared on InformationWeek, “It's one thing to get an insight and quite another to put that insight into action.” It is essential to turn data and analytics into actions that will make the enterprise grow and differentiate. One way of doing this is through nudging. The nudge theory was developed by economist Richard Thaler in 2008 but only rewarded with the Nobel prize last October 2017. Using analytics and a smart user experience, enterprises can nudge or steer employees and consumers into make specific decisions aligned with their goals. In some ways, this is similar to getting a Google Maps input at your next intersection – a nudge – that crystalizes significant analytics in the background around destination routing and anticipated traffic and distills it down to a simple nudge: turn left.

Let’s take another example of an effective analytics strategy put into action in wealth management. Say a client sent an email three years ago requesting to postpone an appointment because of a conflict with his son’s middle school graduation. This email can sit in the firm’s archives with millions of other emails. Three years later, with analytics and NLP, the information from this email can be used to prompt the financial advisor to contact the client and suggest a car insurance for the son, since he is now of driving age. The smart use of analytics provides the predictive patterns for the client’s needs. But the nudge to the right financial advisor at the right time, with the right contextual information, is what makes it come to life.

In summary, to fully reap the benefits and cash in on the value of data, it is essential for business leaders to start with the right vision of driving their business on the back of analytics. Then, build strong competencies in data strategy and architecture, data engineering, data sciences, and machine learning. Finally, use domain knowledge and process insights to turn these insights into action.

This article is published as part of the IDG Contributor Network. Want to Join?

NEW! Download the Spring 2018 digital edition of CIO magazine