Business Intelligence: Enterprise Reporting (Part X: Reports’ Lifecycle)

Introduction

A report’s lifecycle is the sequence of stages through which a report goes during the timespan of its ownership. The main stages resume mainly to report’s definition, development, testing and deployment, however a report’s life occurs within the context of IT processes like Change, Incident/Problem, Access, Availability, Information Security and Knowledge Management. To them can add up Data Management processes like Data Governance, Data Quality and Metadata Management. Therefore, the extended reports’ lifecycle could take the following form:

The processes can be easily tailored to an organization’s needs, even if it may take several attempts until the best mix is found. The activities introduced by the supporting processes don’t necessarily change the way reports are developed as long the processes integrate smoothing in report’s authoring.

Definition Phase

The lifecycle of a report starts with a series of steps that lead to report’s definition and the requirements associated with it:

The starting point is the identification of a need for data. It can be a business question that needs to be answered, a decision that needs to be made, data needed to keep an operational, tactical or strategical objective under control, and so on. Such business situations can be referred simple as (business) problems.

Problem definition

Problem definition (statement) is the process by which a business issue or need is clearly and concisely stated. This step might seem trivial and implied, however in praxis correlated to it lies the most important volume of overwork.

The dictum “a problem well stated is a problem half-solved” applies as well in BI field. Unfortunately, there are cases in which the users want something else than stated or they leave important details out. Sometimes the users aren’t sure what they need/want, and it comes in developer’s attributions to help clarify the problem and put it within a context.

There are cases in which the users just request a report without specifying the problem they need to solve. This might do when the user has a good understanding of the data and the problem, however this approach does not always work. Personally, I find it useful to define for each report also the underneath problem. I see it as a “win-win” situation in which the user invests some knowledge into the developer and thus the developer will better understand the business, while in time he can provide better help. A thorough understanding of the business and knowledge of the users and their needs can help minimize the volume of overwork involved in reports’ development.

Requirements definition

Requirements definition is the process by which functional and non-functional expectations, targets and specifications are elicited and documented.

Functional requirements specify what the report must do - how the report is structured or formatted, how data need to be visualized or navigated, to what file formats need to be exported, on whether needs to be printed, how the data needs to be grouped, in which order, in what currency/language needs to be displayed, what data sources need to be used, etc. The functional requirements are typically listed in the use case and test script.

Non-functional requirements refer to requirements related to report’s accessibility, availability, performance, compliance, documentation, quality, maintainability, security or testability.

The degree to which a requirement can be fulfilled depends entirely on the reporting platform. It can be differentiated between soft and hard constraints. Soft constraints can be overcome by adding more processing power, memory or other types of resources, while hard constraints can’t be easily or at all overcome. Of course, not all requirements are equally important. Important not fulfilled requirements can make a report unusable and, in extremis, can lead to choosing one reporting platform over another.

The requirements can be elicited by a developer, an analyst/consultant or defined by the business itself. Organizations can simplify the process by defining a set of guidelines and standards that need to be considered in reports’ definition. Normally, is enough to reference the document(s) where the guidelines and standards are found. In contrast to other software artifacts, the requirements for reports can be gather in a simplified version of a document. Quite often a checklist can help identify these requirements upfront with a minimum of overhead.

Report definition

Report definition is the process by which report’s content, logic and layout are explicitly defined - what attributes are needed for output and from what source, what static/dynamic parameters are needed, how the data need to be displayed/formatted, what formulas, aggregations or ordering apply.

A report’s definition can be anything between a simple statement summarizing what the report is about and complex structures (mainly in form of a mapping) reflecting in detail each attribute, constraint, formula, grouping or sorting.

A good definition should allow a developer to create the report as needed by the users, eventually with minimal deviations implied by user’s understanding. The holy grail in report’s definition is finding a structure flexible enough to cover all the aspects of a report. Even if some structures allow such flexibility, sometimes it’s almost impossible not provide additional descriptions in textual forms. The less insight the developer has into the business, the more textual descriptions and visuals are needed to be included to support the knowledge gap.

GAP Analysis

GAP Analysis is the iterative process by which the current state of a software artifact or situation is compared with the potential or desired state. It became an integrant tool from professionals’ thinking to the extent its role as separate process is quite often ignored. In the context of reporting authoring it can be used when comparing the requirements against the current infrastructure and the data available, as well while comparing the developed report against the requirements.

It can happen that the technical and data constraints don’t allow building the report as needed by the users. The differences need to be mitigated and eventually the requirements need to be changed to accommodate the reality. In extremis must be considered whether the report still make sense in the light of the modified requirements.

Solution formulation

Solution formulation is the process by which a formal (technical) solution is defined for the given requirements. It’s a conceptualization (aka concept) of the requirements, and in many cases it’s just a short description by which means the report will be build and what data sources will be used. In more complex cases it can include details about the changes needed in the infrastructure to support the report (e.g. creation/extensions of tables and other database objects, ETL jobs, components, etc.), about the data that need to be collected, etc.

Of course, the conceptualization must be considered together with report’s definition. In fact, report’s definition can be considered as part of the conceptualization. A conceptualization can cover multiple reports, as well two or more different solutions can be provided for different sets of reports. The infrastructure can make a concept futile, either when there is a single reporting platform, or when clear rules are in place.

Prototyping

Prototyping is the iterative process of building a simplified version of the report for demonstration and evaluation purposes, so that users can better define the requirements or to prove the concept. The prototype is a preliminary version that can be refined successively until user’s requirements have a final form. It can take the form of a mock-up query to verify report’s technical and logical feasibility, and/or an Excel layout to depict how the report will look like. Prototypes can facilitate the communication between the parties involved and can be considered as part of the requirements.

A prototype might be needed 1 from 5 cases or so, however this number depends also on the number of queries available or of the knowledge of the source and business processes. Because a prototype can involve additional work, it’s important to identify those cases in which a prototype makes sense and keep the effort to a minimum, especially when an approval is involved in the process. Therefore, one should consider the most important characteristics that need to be proved (e.g. if the data can be aggregated, matched, displayed at the requested level of detail, or in the requested format).

With the help of self-service tools, the business has the capabilities to play with the data and find answers by itself, being able thus to create a prototyped version of the report. Once the report met business needs it can be standardized so it can be used organization-wide. It’s recommended to standardize the reports that are used as part of organization’s processes, otherwise self-service can become a bottleneck for the organization.

Change Management

Change Management is the process of ensuring that the changes performed to a system, in this case a BI tool or the whole BI infrastructure, are performed with minimal disruption for the business and that risks are kept under control. Changes can be requested via standard requests or change requests. A standard request (SR) is a pre-approved change that involves low risks, is relatively common and follows a predefined procedure. In contrast to SRs, a change request (CR) requires the authorization of a board, e.g. the Change Advisory Board (CAB), it often involves risks, an investment and the approach is not that common.

Both are hard-copy or electronic templates that allow to capture information about the changes and allow to document the change and track its status. They include typically the problem definition together with users’ requirements, report definition and the formulation of the solution. What differentiates them thus is the approval process that can be sometimes time-consuming, and the volume of formalism needed to manage the requests (e.g. tracking status, writing status reports, handling risks, etc.).

Unless infrastructural changes are necessary, the risks involved with the creation of reports are relatively small, especially when the reports are developed in-house. Reports developed by vendors involve more risks and imply investments that in a form or other need to be approved. Considering the particularities of the two approaches, personally I think that reports that can be developed with internal resources should be done via SRs, while reports developed externally should be done via CRs. Even if this categorization has the potential of creating some confusion, the use of SRs allows reducing the volume of effort necessary to manage the requests. I suppose there can be found solutions to request external changes via SRs as well (e.g. by using contingents and a set of well-defined rules).