Google Business Intelligence key terms

The practice of collecting, storing, and analyzing business data to support planning and decision-making. It combines people, processes, and tools that turn raw records into information leaders can act on.

Choosing a course of action based on evidence from data rather than intuition or seniority alone. It requires trustworthy data, clear questions, and a habit of testing assumptions before committing resources.

A measurable value that shows how effectively a team or organization is achieving a specific objective. Good KPIs are limited in number, tied to strategy, and reported consistently over time.

A quantifiable measure used to track a process or outcome, such as revenue per user or average handling time. Metrics become meaningful when paired with a definition, an owner, and a comparison point.

Any person who uses, funds, or is affected by a BI product, including executives, analysts, and operational staff. Identifying stakeholders early shapes the questions, scope, and format of the final deliverable.

The process of interviewing users and documenting what a report must answer, for whom, and how often. It captures business questions, constraints, and success criteria before any data work begins.

A specific, answerable question about the organization that analysis is meant to resolve, such as which region lost the most customers last quarter. It frames scope and prevents aimless exploration.

Analysis that summarizes what has already happened, using totals, trends, and comparisons. It forms the base layer of reporting and answers questions about past performance rather than future likelihood.

Analysis aimed at explaining why an observed outcome occurred, often by segmenting data and testing relationships. It goes beyond summary totals to isolate drivers such as seasonality, pricing, or channel mix.

Techniques that estimate future or unknown outcomes from historical patterns, such as forecasting demand or scoring churn risk. Results are probabilities and must be validated before driving decisions.

Open-ended investigation of a dataset to find patterns, outliers, and relationships before formal modeling or reporting. It typically uses quick summaries and charts rather than fixed dashboards.

The ability to read, interpret, question, and communicate with data. It includes recognizing misleading charts, understanding basic statistics, and knowing when a number cannot be trusted.

The policies, roles, and standards that control who can access, change, and share data. It protects accuracy and privacy while making data easier to find and reuse across teams.

The degree to which data is accurate, complete, consistent, and timely for its intended use. Poor quality usually surfaces as duplicate records, missing values, or mismatched formats across sources.

Examining a dataset's structure and content to learn value ranges, null rates, distinct counts, and formats. Profiling reveals problems early, before transformations and models are built on top of the data.

A structured representation of how data entities relate to one another, such as customers, orders, and products. It defines tables, keys, and relationships so queries return consistent results.

A data store organized into tables of rows and columns that are linked by shared keys. It enforces structure through schemas and is queried with SQL.

A column or set of columns whose value uniquely identifies each row in a table. It cannot contain nulls and prevents duplicate records from being inserted.

A column that references the primary key of another table, creating a link between the two. It enforces referential integrity so child records cannot point to nonexistent parents.

The defined structure of a database, including its tables, columns, data types, and constraints. In analytics, it also describes how organized layers of data are arranged for consumption.

Organizing tables to reduce redundancy by separating related facts into distinct tables. It improves consistency for transactional systems but often requires joins that slow analytical queries.

Deliberately combining tables or duplicating attributes to reduce joins and speed up reads. It is common in analytical stores where query performance matters more than storage efficiency.

A modeling pattern with a central fact table surrounded by dimension tables linked by keys. It simplifies queries and is widely used in data warehouses and BI semantic layers.

A table storing measurable events, such as sales transactions, along with keys to related dimensions. It usually holds numeric measures and grows quickly because it records each occurrence.

A table holding descriptive attributes used to filter and group facts, such as product category, region, or customer segment. It is typically smaller than the fact table and joined by a surrogate key.

An attribute used to slice or categorize a measure, such as date, channel, or geography. Dimensions answer the 'by what' part of a question while measures answer the 'how much' part.

A visual map of entities, their attributes, and the relationships between them, often showing cardinality such as one-to-many. It helps teams agree on structure before building tables.

A set of automated steps that moves data from sources to a destination, applying extraction, transformation, and loading along the way. Reliable pipelines run on a schedule and handle failures gracefully.

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

Extract, Load, Transform: raw data is loaded into a warehouse first and transformed there using its compute power. This suits cloud warehouses where storage is cheap and SQL transformation is flexible.

A centralized repository optimized for analytical queries, holding integrated data from many sources. It typically stores modeled, historical data used for reporting rather than live transactions.

A storage layer that holds raw data in its native format, including files, logs, and unstructured content. It offers flexibility but requires governance to avoid becoming an unusable dump.

Correcting or removing inaccurate, duplicated, or malformed records before analysis. Typical tasks include standardizing formats, handling nulls, trimming whitespace, and resolving conflicting values.

An operation that combines rows from two tables based on a related column. Inner, left, right, and full joins differ in which unmatched rows are kept, which directly changes result counts.

Collapsing many rows into summary values using functions such as sum, count, average, or maximum. It is usually paired with grouping so each category receives its own summary.

A visual display of the most important metrics for a specific audience, arranged so status can be read at a glance. It updates from live or scheduled data and usually allows limited filtering.

A structured, often static presentation of data designed to answer specific questions or document results for a period. Reports favor completeness and detail over at-a-glance monitoring.

Encoding data as charts, maps, or other graphics so patterns and comparisons are easier to perceive. Effective choices match the chart type to the question and avoid unnecessary decoration.

A control or condition that limits which records appear in a view, such as a date range or region selector. Good filters are obvious to users and never silently hide important data.

Navigation that lets a user move from a summary level to finer detail, for example from yearly revenue to individual transactions. It supports both quick scanning and deeper investigation in one tool.

A structured effort to identify the underlying reason a problem occurred rather than only its symptom. In BI it often means segmenting metrics and comparing periods to isolate the true driver.

Presenting analysis as a clear narrative that connects evidence to a recommended action. It pairs a strong headline and visuals with context about uncertainty and next steps.