23 Sep How Analytics Has Changed in the Last 10 Years: 5 Ways YourData Strategy Can Fail
In your own words, post a ONE PARAGRAPH REVIEW on one of the following HBR articles listed below. No need for resources.
NO PLAGIARISM!!!!!!!!!
What’s Your Data Strategy?
From the May–June 2017 Issue
· Summary
· Save
· Share
· Comment
Having a CDO and a data-management function is a start, but neither can be fully effective in the absence of a coherent strategy for organizing, governing, analyzing, and deploying an organization’s information assets. Indeed, without such strategic management many companies struggle to protect and leverage their data—and CDOs’ tenures are often difficult and short (just 2.4 years on average, according to Gartner). In this article we describe a new framework for building a robust data strategy that can be applied across industries and levels of data maturity. The framework draws on our implementation experience at the global insurer AIG (where DalleMule is the CDO) and our study of half a dozen other large companies where its elements have been applied. The strategy enables superior data management and analytics—essential capabilities that support managerial decision making and ultimately enhance financial performance.
The “plumbing” aspects of data management may not be as sexy as the predictive models and colorful dashboards they produce, but they’re vital to high performance. As such, they’re not just the concern of the CIO and the CDO; ensuring smart data management is the responsibility of all C-suite executives, starting with the CEO.
Defense Versus Offense
Our framework addresses two key issues: It helps companies clarify the primary purpose of their data, and it guides them in strategic data management. Unlike other approaches we’ve seen, ours requires companies to make considered trade-offs between “defensive” and “offensive” uses of data and between control and flexibility in its use, as we describe below. Although information on enterprise data management is abundant, much of it is technical and focused on governance, best practices, tools, and the like. Few if any data-management frameworks are as business-focused as ours: It not only promotes the efficient use of data and allocation of resources but also helps companies design their data-management activities to support their overall strategy.
Data defense and offense are differentiated by distinct business objectives and the activities designed to address them. Data defense is about minimizing downside risk. Activities include ensuring compliance with regulations (such as rules governing data privacy and the integrity of financial reports), using analytics to detect and limit fraud, and building systems to prevent theft. Defensive efforts also ensure the integrity of data flowing through a company’s internal systems by identifying, standardizing, and governing authoritative data sources, such as fundamental customer and supplier information or sales data, in a “single source of truth.” Data offense focuses on supporting business objectives such as increasing revenue, profitability, and customer satisfaction. It typically includes activities that generate customer insights (data analysis and modeling, for example) or integrate disparate customer and market data to support managerial decision making through, for instance, interactive dashboards.
Offensive activities tend to be most relevant for customer-focused business functions such as sales and marketing and are often more real-time than is defensive work, with its concentration on legal, financial, compliance, and IT concerns. (An exception would be data fraud protection, in which seconds count and real-time analytics smarts are critical.) Every company needs both offense and defense to succeed, but getting the balance right is tricky. In every organization we’ve talked with, the two compete fiercely for finite resources, funding, and people. As we shall see, putting equal emphasis on the two is optimal for some companies. But for many others it’s wiser to favor one or the other.
Some company or environmental factors may influence the direction of data strategy: Strong regulation in an industry (financial services or health care, for example) would move the organization toward defense; strong competition for customers would shift it toward offense. The challenge for CDOs and the rest of the C-suite is to establish the appropriate trade-offs between defense and offense and to ensure the best balance in support of the company’s overall strategy.
Decisions about these trade-offs are rooted in the fundamental dichotomy between standardizing data and keeping it more flexible. The more uniform data is, the easier it becomes to execute defensive processes, such as complying with regulatory requirements and implementing data-access controls. The more flexible data is—that is, the more readily it can be transformed or interpreted to meet specific business needs—the more useful it is in offense. Balancing offense and defense, then, requires balancing data control and flexibility, as we will describe.
Single Source, Multiple Versions
Before we explore the framework, it’s important to distinguish between information and data and to differentiate information architecture from data architecture. According to Peter Drucker, information is “data endowed with relevance and purpose.” Raw data, such as customer retention rates, sales figures, and supply costs, is of limited value until it has been integrated with other data and transformed into information that can guide decision making. Sales figures put into a historical or a market context suddenly have meaning—they may be climbing or falling relative to benchmarks or in response to a specific strategy.
A company’s data architecture describes how data is collected, stored, transformed, distributed, and consumed. It includes the rules governing structured formats, such as databases and file systems, and the systems for connecting data with the business processes that consume it. Information architecture governs the processes and rules that convert data into useful information. For example, data architecture might feed raw daily advertising and sales data into information architecture systems, such as marketing dashboards, where it is integrated and analyzed to reveal relationships between ad spend and sales by channel and region.
Many organizations have attempted to create highly centralized, control-oriented approaches to data and information architectures. Previously known as information engineering and now as master data management, these top-down approaches are often not well suited to supporting a broad data strategy. Although they are effective for standardizing enterprise data, they can inhibit flexibility, making it harder to customize data or transform it into information that can be applied strategically. In our experience, a more flexible and realistic approach to data and information architectures involves both a single source of truth (SSOT) and multiple versions of the truth (MVOTs). The SSOT works at the data level; MVOTs support the management of information.
In the organizations we’ve studied, the concept of a single version of truth—for example, one inviolable primary source of revenue data—is fully grasped and accepted by IT and across the business. However, the idea that a single source can feed multiple versions of the truth (such as revenue figures that differ according to users’ needs) is not well understood, commonly articulated, or, in general, properly executed.
The key innovation of our framework is this: It requires flexible data and information architectures that permit both single and multiple versions of the truth to support a defensive-offensive approach to data strategy.
The Elements of Data Strategy
|
DEFENSE |
OFFENSE |
|
|
KEY OBJECTIVES |
Ensure data security, privacy, integrity, quality, regulatory compliance, and governance |
Improve competitive position and profitability |
|
CORE ACTIVITIES |
Optimize data extraction, standardization, storage, and access |
Optimize data analytics, modeling, visualization, transformation, and enrichment |
|
DATA-MANAGEMENT ORIENTATION |
Control |
Flexibility |
|
ENABLING ARCHITECTURE |
SSOT (Single source of truth) |
MVOTs (Multiple versions of the truth) |
|
From “WHAT’S YOUR DATA STRATEGY?” BY LEANDRO DALLEMULE AND THOMAS H. DAVENPORT, MAY–JUNE 2017 |
© HBR.ORG |
Find this and other HBR graphics in our Visual Library
OK. Let’s parse that.
The SSOT is a logical, often virtual and cloud-based repository that contains one authoritative copy of all crucial data, such as customer, supplier, and product details. It must have robust data provenance and governance controls to ensure that the data can be relied on in defensive and offensive activities, and it must use a common language—not one that is specific to a particular business unit or function. Thus, for example, revenue is reported, customers are defined, and products are classified in a single, unchanging, agreed-upon way within the SSOT.
Not having an SSOT can lead to chaos. One large industrial company we studied had more than a dozen data sources containing similar supplier information, such as name and address. But the content was slightly different in each source. For example, one source identified a supplier as Acme; another called it Acme, Inc.; and a third labeled it ACME Corp. Meanwhile, various functions within the company were relying on differing data sources; often the functions weren’t even aware that alternative sources existed. Human beings might be able to untangle such problems (though it would be labor-intensive), but traditional IT systems can’t, so the company couldn’t truly understand its relationship with the supplier. Fortunately, artificial intelligence tools that can sift through such data chaos to assemble an SSOT are becoming available. The industrial company ultimately tapped one and saved substantial IT costs by shutting down redundant systems. The SSOT allowed managers to identify suppliers that were selling to multiple business units within the company and to negotiate discounts. In the first year, having an SSOT yielded $75 million in benefits.
A New Data Architecture Can Pay for Itself
When companies lack a robust SSOT-MVOTs data architecture, teams across the organization may create and store the data they need in siloed repositories that vary in depth, breadth, and formatting. Their data management is often done in isolation with inconsistent requirements. The process is inefficient and expensive and can result in the proliferation of multiple uncontrolled versions of the truth that aren’t effectively reused. Because SSOTs and MVOTs concentrate, standardize, and streamline data-sourcing activities, they can dramatically cut operational costs.
One large financial services company doing business in more than 200 countries consolidated nearly 130 authoritative data sources, with trillions of records, into an SSOT. This allowed the company to rationalize its key data systems; eliminate much supporting IT infrastructure, such as databases and servers; and cut operating expenses by automating previously manual data consolidation. The automation alone yielded a 190% return on investment with a two-year payback time. Many companies will find that they can fund their entire data management programs, including staff salaries and technology costs, from the savings realized by consolidating data sources and decommissioning legacy systems.
The CDO and the data-management function should be fully responsible for building and operating the SSOT structure and using the savings it generates to fund the company’s data program. Most important is to ensure at the outset that the SSOT addresses broad, high-priority business needs, such as applications that benefit customers or generate revenue, so that the project quickly yields results and savings—which encourages organization-wide buy-in.
Read more
An SSOT is the source from which multiple versions of the truth are developed. MVOTs result from the business-specific transformation of data into information—data imbued with “relevance and purpose.” Thus, as various groups within units or functions transform, label, and report data, they create distinct, controlled versions of the truth that, when queried, yield consistent, customized responses according to the groups’ predetermined requirements.
Consider how a supplier might classify its clients Bayer and Apple according to industry. At the SSOT level these companies belong, respectively, to chemicals/pharmaceuticals and consumer electronics, and all data about the supplier’s relationship with them, such as commercial transactions and market information, would be mapped accordingly. In the absence of MVOTs, the same would be true for all organizational purposes. But such broad industry classifications may be of little use to sales, for example, where a more practical version of the truth would classify Apple as a mobile phone or a laptop company, depending on which division sales was interacting with. Similarly, Bayer might be more usefully classified as a drug or a pesticide company for the purposes of competitive analysis. In short, multiple versions of the truth, derived from a common SSOT, support superior decision making.
A company’s position on the offense-defense spectrum is rarely static.
At a global asset management company we studied, the marketing and finance departments both produced monthly reports on television ad spending—MVOTs derived from a common SSOT. Marketing, interested in analyzing advertising effectiveness, reported on spending after ads had aired. Finance, focusing on cash flow, captured spending when invoices were paid. The reports therefore contained different numbers, but each represented an accurate version of the truth.
Procter & Gamble has adopted a similar approach to data management. The company long had a centralized SSOT for all product and customer data, and other versions of data weren’t allowed. But CDO Guy Peri and his team realized that the various business units had valid needs for customized interpretations of the data. The units are now permitted to create controlled data transformations for reporting that can be reliably mapped back to the SSOT. Thus the MVOTs diverge from the SSOT in consistent ways, and their provenance is clear.
image1.png
,
How Analytics Has Changed in the Last 10 Years (and How It’s Stayed the Same)
June 22, 2017
· Summary
· Save
· Share
· Comment
·
Blockchain: Tools for Preparing Your Team for the Future
Book
49.95 View Details
·
Clean Edge Razor: Splitting Hairs in Product Positioning
HBS Brief Case
8.95 View Details
·
Deutsche Allgemeinversicherung
Photo by Ferdinand Stöhr
Ten years ago, Jeanne Harris and I published the book Competing on Analytics , and we’ve just finished updating it for publication in September. One major reason for the update is that analytical technology has changed dramatically over the last decade; the sections we wrote on those topics have become woefully out of date. So revising our book offered us a chance to take stock of 10 years of change in analytics.
Of course, not everything is different. Some technologies from a decade ago are still in broad use, and I’ll describe them here too. There has been even more stability in analytical leadership, change management, and culture, and in many cases those remain the toughest problems to address. But we’re here to talk about technology. Here’s a brief summary of what’s changed in the past decade.
The last decade, of course, was the era of big data. New data sources such as online clickstreams required a variety of new hardware offerings on premise and in the cloud, primarily involving distributed computing — spreading analytical calculations across multiple commodity servers — or specialized data appliances. Such machines often analyze data “in memory,” which can dramatically accelerate times-to-answer. Cloud-based analytics made it possible for organizations to acquire massive amounts of computing power for short periods at low cost. Even small businesses could get in on the act, and big companies began using these tools not just for big data but also for traditional small, structured data.
Insight Center
Analytics are critical to companies’ performance.
Along with the hardware advances, the need to store and process big data in new ways led to a whole constellation of open source software, such as Hadoop and scripting languages. Hadoop is used to store and do basic processing on big data, and it’s typically more than an order of magnitude cheaper than a data warehouse for similar volumes of data. Today many organizations are employing Hadoop-based data lakes to store different types of data in their original formats until they need to be structured and analyzed.
Since much of big data is relatively unstructured, data scientists created ways to make it structured and ready for statistical analysis, with new (and old) scripting languages like Pig, Hive, and Python. More-specialized open source tools, such as Spark for streaming data and R for statistics, have also gained substantial popularity. The process of acquiring and using open source software is a major change in itself for established businesses.
The technologies I’ve mentioned for analytics thus far are primarily separate from other types of systems, but many organizations today want and need to integrate analytics with their production applications. They might draw from CRM systems to evaluate the lifetime value of a customer, for example, or optimize pricing based on supply chain systems about available inventory. In order to integrate with these systems, a component-based or “microservices” approach to analytical technology can be very helpful. This involves small bits of code or an API call being embedded into a system to deliver a small, contained analytical result; open source software has abetted this trend.
Related Video
The Explainer: Big Data and Analytics
<span>What the two terms really mean — and how to effectively use each.</span>
· Save
· Share
This embedded approach is now used to facilitate “analytics at the edge” or “
