Information is critical to managing business change. Rapidly reacting to business events and proactive business planning requires accurate and timely information. Building a performance management solution would mean constantly managing corporate data with the right portfolio of tools and technologies. Understanding and analyzing corporate data becomes easier within the context of an enterprise information management framework.Data discovery acts as a bridge between information management and performance management. This needs a thorough understanding of all business processes, alignment of corporate strategies with data and seamless integration of performance analysis techniques. This article discusses a structured approach for data discovery for implementation of performance management solutions.

Data Discovery in Information Management

Data discovery assists in managing such business changes by providing a comprehensive inventory or taxonomy of corporate data. The outputs and models generated from such an exercise can be used for any enterprise-wide solution that requires comprehensive information management.

Business change translates to changes in corporate strategies and plans, modification of existing key performance indicators (KPIs) and analysis of newer KPIs. Nevertheless, the fundamental data elements required for this change in performance analysis generally remain stable. The patterns, calculations or indicators are frequently modified to make performance management more dynamic and to effectively monitor performance levels.

Knowing Your Data

This paper discusses a detailed methodology for the data discovery phase of any performance management or a business intelligence (BI) solution. The seven subphases of the methodology provide an exhaustive checklist to uncover the corporate data across the enterprise.

Figure 1: Seven Steps to Data Discovery

Each step progressively captures the required attributes of corporate data. The various attributes of data are collected with the help of interviews that are conducted with business and technology users like data source owners and database administrators. The interviews in the data discovery phase impact architectural decisions of the performance management solution.

Consider the following example illustrating data discovery for the Customer acquisition KPI. A structured approach helps in capturing details on all aspects of the KPI.

Figure 2: Data Discovery of Customer Acquisition KPI

Understanding Business Systems and Processes

This is the first step to knowing your data. Pick each process and understand the flow of data across the process. Such a business process map or context diagram of the enterprise-wide systems and processes helps understand the interdependencies between them. Also, they indicate the direction of the flow of data, the data sources and applications in the system, and users. They highlight the actions and agents involved in the process. E.g. stocks are reordered when the inventory is below 100 items. Goods are ordered by the procurement supervisor and approved by his/her manager.

Tools and standardization techniques are available that can be used to capture such process details as the Unified Modeling Language (UML).

Listing Key Performance Indicators

After understanding the complete picture of the systems and processes, identify all KPIs for each of the above processes. Identifying KPIs is one of key activities of any corporate performance management (CPM) exercise.

Prioritize the list for the current assessment period and group them into major business segments or perspectives like customer, finance, learning and growth, and business process. The complete KPI list could be extensive but prioritize the KPIs that matter to the organization. These KPIs should align with the corporate strategies and objectives.

Subsequently, capture the relevant details of this prioritized set of KPIs. The details should also include the dimensions across which the KPIs will be analyzed and the access permissions to be provided to users for the solution objects and components.

Mapping KPIs to Metrics

After listing down the KPIs and their details, all KPIs should be further drilled down into its elementary constituents called metrics. KPI's consists of multiple metrics, and each metric consists of elementary data elements. In complex KPI calculations, you might be required to traverse several levels before you get to the basic elementary metric. In some cases, the KPI's could simply consist of data elements.

Metrics are independent analytical numbers, but KPIs have to be compared with targets and are benchmarked to measure performance. Break down the KPI's into metrics and capture all the relevant details of the metrics iteratively for all the levels in the metric tree.

Defining Threshold Values

Benchmarking helps in proactive analytics and performance monitoring. Benchmarks quantify and provide insight into whether the performance of the KPI has been improving or dropping. Any KPI is incomplete without its benchmarks and thresholds.

Business intelligence (BI) systems consist of numerous mechanisms for information delivery, e.g., graphs, charts, alerts, triggers, traffic lights and dials. These channels of information delivery need special data called benchmarks or thresholds. As an integral part of any discovery process, benchmarks and threshold values for information delivered across various components and channels must be captured.

The values could either be an absolute value or a range of values. These values are indicated by red, green, yellow flags. The threshold value can be derived or extracted automatically from a rule repository. In case of a manual update, capture the user or user group who will maintain these thresholds. Also, identify if it's local or global by understanding if the value can be used only for a particular division or across the organization.

Mapping Metrics to Data Elements

Metrics further consists of individual data elements coming from various data sources. In this process, break down the metrics into data elements and capture all their relevant details. This phase is commonly called the metric-mapping phase.

With the data source details that would include the owner and steward details and frequency of availability, also capture the granular details of metrics, data elements and its attributes. The data elements can be grouped and captured for each data source.

Eliminating Data Redundancy

Some of the general problems that organizations face these days are lack of dual or overlapping ownership of data. The same data element can be sourced from multiple data sources. This redundancy could help in reconciling the data but could also raise a conflict in ownership issues. The data could be entered in the systems via different channels, but there should be a single value that can be considered as accurate and final. There should be a single department or individual who owns the data.

Using governance policies and reconciliation rules, every conflicting data element should be validated and arrived at by a common consensus. While discovering data for such data elements, try to define a system of record between conflicting data elements. Resolving the problem of the system of record for all data elements becomes time-consuming but is critical for projects involving data integration from multiple sources.

Applying Business Rules and Calculations

Once the attributes of the data elements have been discovered and documented, the data can be integrated from the data sources. The calculations required for the KPI's or modifications required for data elements should be centralized.

The changes or calculation applied to the KPIs and data elements are called transformations. During the process of data extraction from the data sources, the transformations are applied to the data elements. Capture the details that help derive the right data integration strategy and architecture, like the functional extraction rules and the source to target mapping, with any validation rules (if any).

The frequency of the data integration depends on the frequency of need as captured in the earlier steps mentioned in the data discovery methodology. Real-time data integration must be rightly justified as costs and time required for implementation and support of the integration process can be high.

With the integration and availability of corporate-wide data, organizations can make informed decisions and can also manage business dynamics. Only after bringing data under a common framework can it be utilized for monitoring and analyzing corporate performance.

A successful data discovery exercise is vital to setting up an infrastructure for information management to support enterprise-wide solutions for performance management and business intelligence. Defining the right granularity of data elements, de-linking KPI definitions and calculations from data, defining and identifying common data across initiatives, as described in this paper, help uncover the treasure-trove of corporate information. A structured approach towards data discovery helps in optimally cataloging and leveraging corporate data for business competitiveness.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access