EXIL-CBI key terms

The practice of collecting, integrating, and presenting an organization's data so that people can monitor performance and make informed decisions. It combines technology, processes, and people rather than being a single software product.

A small set of measures chosen to show whether an organization is achieving a specific strategic objective. Each KPI has a defined target, an owner, and a reporting cadence so progress can be judged objectively.

A quantifiable value calculated from data that describes some aspect of performance, such as revenue, volume, or elapsed time. Metrics become indicators only when compared against a target or a trend.

The use of historical data to explain what has already happened, typically through summaries, trends, and comparisons. It answers what occurred rather than what will occur or what action should be taken.

Techniques that use historical data and statistical or machine-learning models to estimate the likelihood of future outcomes. Results are probabilities and ranges rather than certainties, so they must be interpreted with care.

A centralized repository that consolidates data from many source systems into a consistent, query-optimized structure used for reporting and analysis. It typically holds historical, integrated, subject-oriented data separated from day-to-day transaction processing.

A subset of a data warehouse, or a small warehouse in its own right, built for a specific department, subject area, or audience. Marts simplify access and can be dependent on or independent of the central warehouse.

A storage environment that holds large volumes of raw data in its native format until it is needed, often using low-cost object storage. Structure is applied at read time rather than enforced on write.

Extract, Transform, Load: a data integration pattern in which data is pulled from sources, cleaned and reshaped in a separate processing engine, then written to the target store. Transformation happens before loading.

Extract, Load, Transform: a pattern in which raw data is loaded into the target platform first and transformed afterwards using that platform's own compute power. It suits scalable, cloud-based analytical platforms.

Online Analytical Processing: a way of organizing and querying data as multidimensional structures so users can slice, dice, and drill through it quickly. It is optimized for summarizing large numbers of records rather than for recording transactions.

The central table in a dimensional model that stores numeric measurements of business events plus foreign keys linking to the relevant dimensions. It is usually long and narrow, with one row per event at a defined grain.

A table that holds the descriptive attributes used to filter, group, and label the numbers in a fact table, such as product, customer, or date details. It provides the context by which analysis is performed.

A numeric field in a fact table that can be summed, averaged, or otherwise aggregated, such as quantity sold or amount billed. Non-additive measures require care because totals across some dimensions may be meaningless.

A design technique that organizes data into facts, which record events and their numbers, and dimensions, which describe them. It prioritizes understandability and query performance for business users over strict normalization.

A dimensional design with a single fact table joined directly to a set of denormalized dimension tables, resembling a star shape. It keeps queries simple and fast by avoiding long chains of joins.

The level of detail represented by one row of a fact table, stated precisely, for example one row per order line per day. Declaring grain first prevents mixed or ambiguous measures.

A dimension design that preserves history when attribute values change, such as a customer's address. Common approaches add a new row per change with effective dates, or keep only the current value.

A system-generated, meaningless integer assigned to each row of a dimension or fact table to identify it uniquely. It insulates the warehouse from changes in source-system natural keys.

A defined set of business-friendly names, calculations, relationships, and rules that sits between raw data and the tools users query it with. It keeps metrics consistent across reports and applications.

A single screen that presents a focused set of indicators, usually graphically, so a user can assess status at a glance. Good dashboards support a decision rather than displaying every available number.

The representation of data as charts, maps, or other graphics so that patterns and exceptions are easier to see. The choice of visual encoding strongly affects how accurately a viewer reads the data.

The structured process of discovering what stakeholders need from a solution through interviews, workshops, observation, and document review. It produces agreed, testable statements of need rather than assumed solutions.

An automated sequence of steps that moves data from sources through processing to a destination, often on a schedule or in response to events. Pipelines handle extraction, validation, transformation, and loading.

An approach that lets business users build their own queries, reports, and dashboards from governed data with suitable tools, reducing reliance on specialist teams. It depends on curated data and clear definitions to avoid inconsistent results.

Formal testing in which intended users confirm that a delivered solution meets agreed business requirements and is fit for purpose. Sign-off typically gates the transition to live operation.

A limited, time-boxed build that demonstrates whether a proposed approach can technically work and deliver value. It reduces uncertainty before larger investment but is not itself a production solution.

The decision rights and accountability framework that defines who may take which actions on which data, and under what rules. It covers policy, standards, ownership, and oversight of data as an asset.

A named person accountable for the quality, definitions, and appropriate use of a defined set of data on behalf of the business. Stewards resolve issues and approve changes rather than owning the underlying technology.

A record of where data came from, how it was transformed, and where it flows to across systems and reports. It supports impact analysis, troubleshooting, and audit evidence.

The degree to which data is accurate, complete, consistent, timely, and fit for its intended use. Quality is managed through defined dimensions, rules, measurement, and remediation of underlying causes.

The discipline of creating and maintaining agreed, authoritative records for core entities such as customers, products, and suppliers across the organization. It removes duplication and conflicting versions of the same entity.

An access control mechanism that filters the rows a user can see in a table or report according to rules such as region or department. The same report can therefore show different data to different people.

Replacing or obscuring sensitive values with realistic substitutes so users can work with data without seeing the original information. It is commonly applied to non-production copies and to lower-privilege access.

The controlled process for requesting, assessing, approving, testing, and recording modifications to a production service, so that risk is understood and disruption is minimized. Records of changes also support later investigation.

A documented commitment between a provider and its customers stating the services to be delivered and measurable targets such as availability, response time, and support hours. It usually defines remedies when targets are missed.

A framework describing progressive levels of capability in a discipline, from ad hoc to optimized, with typical characteristics for each level. Organizations use it to assess current state and plan improvement.

A structured investigation that seeks the underlying reason an incident or defect occurred, rather than the immediate symptom. Findings should lead to preventive action so that the problem does not recur.