Data and Business Intelligence Glossary Terms

A test that compares two different versions of a product or interface (typically the existing one and a new variant) by presenting different options at random to different people and gathering data on their behavior.

A program's defined set of endpoints that other programs can interact with.

The annualized value of all currently active recurring subscriptions, usually calculated as MRR multiplied by 12.

Agentic analytics is the use of AI agents that do analytical work in steps — planning, running queries, inspecting results, and revising — not in a single pass.

The act of summarizing data with a mathematical function, such as averaging the values in a column, or counting the number of rows in a table.

A notification about a question that Metabase delivers by email or Slack.

An analytical database is a database optimized for a few large analysis operations rather than many small transactions.

The database in which Metabase stores operation data about questions, dashboards, and other Metabase information.

A type of line chart where the space between plotted values and the x-axis is filled in with a solid color.

An attribute is a property that describes or identifies some entity. In some Metabase plans, user attributes are used to restrict which data people can access.

A record of activity and performance in a Metabase, available to admins in some plans.

An app designed for people to look at data without relying on code.

A data visualization that uses rectangles that are proportional in size to the values they measure.

A single range of continuous values used to group values in a chart.

The grouping of aggregated results by one or more dimensions.

A data visualization that plots data with three variables.

The total sales and marketing cost of acquiring a new customer, divided by the number of new customers acquired in the same period.

A named result set in a SQL query, equivalent to a subquery. CTEs are evaluated each time the main query is run.

A component of a dashboard that displays data or text.

Change data capture (CDC) is a technique for detecting row-level changes in a source database and streaming just those changes to another system.

The share of customers, subscriptions, or recurring revenue you lose over a given period.

COALESCE is a SQL function that takes a list of values and returns the first one that isn't null.

An analysis that groups records by a shared starting event — usually the month they signed up — and tracks each group separately over time.

A set of items in Metabase, including questions, models, dashboards, and other collections.

A list of values, usually belonging to a particular field, displayed vertically in a table.

Columnar storage is a way of laying data out on disk column by column instead of row by row, so analytical queries only read the columns they need.

A chart that combines bars and lines.

A maintenance process that rewrites scattered data — many small files or fragmented pages — into fewer, larger, contiguous units so queries read less and run faster.

A function that combines multiple strings into a single string.

Cross-filtering is clicking on a chart or table in a dashboard to filter everything else in the dashboard.

A chart that shows how numbers accumulate over time and how they are progressing towards a goal.

Sometimes called a "calculated" column. You can use custom expressions to add columns to the results of your query in the notebook editor. For example, adding a column that calculates the difference between the total and subtotal.

A setting on a dashboard card that determines what happens when people click on the card.

A formula in Metabase that uses functions and mathematical operators to allow for more complicated questions, similar to a formula in spreadsheet software.

A directed acyclic graph (DAG) is a graph in which it's impossible to get back to any starting node by following the links between nodes.

A data visualization tool that holds important charts and text, collected and arranged on a single screen.

Dashboard subscriptions allow you to schedule the sending of results of the questions on a dashboard via email or Slack.

A data contract is an explicit, versioned agreement between a dataset's producer and its consumers on schema, meaning, and quality guarantees.

A document that describes the tables, fields, and other elements in a database and explains their meaning and origin.

A data lake is a place to store both structured and unstructured information, typically as files or blobs.

A storage architecture that adds warehouse-style tables, schemas, and transactions on top of cheap object storage, so one copy of your data serves both engineering and analytics.

A data mart is a subset of a data warehouse curated for one team or subject area, along with the tables, models, and dashboards that team uses.

A data mesh is an approach where the teams closest to the data own it, publishing well-defined datasets as products rather than relying on one central team.

Any pattern that organizes and labels information.

A data pipeline is the set of automated steps that move data from where it's produced to where it's analyzed, transforming it along the way.

Metabase's data reference section lets you add and look up metadata about your databases and their tables and fields.

A data sandbox is set of boundaries that define access to a table down to its columns and rows. You can coordinate data sandboxes with your SSO setup.

A field's classification as implemented directly by a database, such as integer, text, or timestamp.

A database used for analytics.

A database index is a data structure the database maintains alongside a table so it can find rows matching a condition without scanning the whole table.

A descriptive, qualitative attribute of your data, like a name, category, or URL field.

To explore a pre-selected subset of data, e.g., by clicking on a bar in a histogram to see the raw values for that bar.

An ERD, or entity relationship diagram, is a graphical representation of how tables in a database connect to each other.

Extract, transform, load: an ETL is a common operation in data processing systems that reads data from sources, massages it, and stores it in another system.

Using third-party software to include charts and dashboards in your application that customers can query.

Placing some functionality of one app inside another. Metabase uses iframes to embed questions, dashboards, or (in some plans) the full Metabase application.

Metabase's field type for the column that uniquely identifies each row in a table — the equivalent of a database primary key.

Sometimes called a column, a field is an attribute of a database table.

Field filters are special variables you can use in SQL queries that let you create "smart" filter widgets.

The equation that powers a custom expression, like the formula you'd enter in the cell of a spreadsheet.

The meaning of a field (or column) in a table.

A filter is a predicate expression that limits the results of a query based on some stated criteria.

A UI element on a dashboard or SQL query where you can input a value or select one from a menu to filter results.

A column in one database table that holds the primary key of a row in another table, creating a link between the two.

A join that keeps every row from both tables regardless of whether keys matched.

A funnel chart is a visualization that shows how a quantity shrinks across the sequential steps of a process, making drop-off between stages easy to spot.

GROUP BY is the SQL clause that collapses rows sharing the same values into groups, so aggregate functions like COUNT and SUM can compute one result per group.

Queries built using Metabase's graphical query builder.

A gauge chart displays a single value on a dial against colored ranges, showing at a glance whether a metric is below, within, or above its target zone.

A popular format for representing maps.

A constant line added as a marker to a chart showing how high (or low) values must be to be considered noteworthy.

A map with values overlaid graphically in a regular grid, e.g., as squared 10km by 10km colored to show values.

HAVING is the SQL clause that filters the results of a GROUP BY query after aggregation, keeping only the groups that meet a condition.

A heatmap uses color intensity to show the magnitude of values across a two-dimensional grid, so patterns and outliers stand out at a glance.

A chart that displays continuous data using vertical bars that each represent a discrete, equal-sized range.

Adding more machines to a system to increase capacity — scaling out rather than up — so work is spread across many servers instead of one bigger one.

A join that only keeps rows where the keys from the input tables matched.

A way to represent data by combining basic values in arrays and key/value structures using the syntactic conventions of JavaScript.

A standard for sharing signed authentication claims between web services.

The combination of results from two tables in a relational database.

A metric that shows progress toward a personal or company goal, and that a team has agreed to be measured by.

A protocol for querying and updating directory services, commonly used as an organization's single source of truth for user accounts, groups, and authentication.

A metric that changes before the outcome it predicts, giving you time to act — as opposed to a lagging indicator, which confirms a result after the fact.

A join that keeps every row from the first (left-hand) table regardless of whether that row matched any in the second (right-hand) table.

A type of visualization that connects discrete values connected by lines to show changes and trends.

An axis in a chart that spaces marks according to their values.

A filter widget that only displays a subset of values depending on the value of another filter.

Hardware or software that sits in front of a group of servers and distributes incoming traffic across them, so no single server is overloaded and failed servers stop receiving requests.

A materialized view stores a query's results as a physical table, so the answer can be read directly instead of recomputed every time.

A numerical attribute of your data that can be broken out by one or more dimensions.

A medallion architecture organizes lakehouse data into layers: raw as it arrived (bronze), cleaned and standardized (silver), aggregated for use (gold).

Information that describes data to make it easier to find, manipulate, and make use of that data.

A metric is a calculation performed on a measure. In Metabase, a Metric is a saved, reusable definition of an aggregation that anyone on your team can use as a starting point for questions.

A fundamental building block in Metabase that you can use as a starting point for new questions.

The Model Context Protocol is an open standard for connecting AI assistants to external tools and data, so any compliant assistant can use any integration.

A software architecture where a single instance or deployment of an application serves multiple, distinct customers, each of which can only see its own data.

A query written as text in the database’s query language (usually SQL).

The editor in Metabase used to write queries as text. It includes a sidebar with three tabs: data reference, SQL variables, and Snippets.

The process of structuring information in a relational database to reduce redundancy.

The single metric a company or team aligns around, chosen because it captures the value customers get from the product.

The GUI interface for asking questions in Metabase, where you can filter, summarize, join tables, and create custom columns.

Online analytical processing: processing that involves a small number of large operations, such as creating quarterly reports.

Online transaction processing: processing that involves a large number of small operations, such as logging user activity on a website.

An axis in a chart that displays values according to the discrete groups those values are in.

The root collection in Metabase.

A special type of variable that specifies an input to a query.

Your semi-private collection in Metabase, viewable only by you and Admins. Your personal collection is a great place to draft questions and dashboards.

A pie chart is a circular visualization that divides a whole into slices, with each slice's size proportional to its share of the total.

A map that displays a single discrete marker for each point of interest.

A data visualization that summarizes rows and columns of a table and lets you rotate (pivot) the columns.

An expression that evaluates to either true or false, like quantity > 0. True and false values are known as Boolean values.

A column (or set of columns) whose value uniquely identifies every row in a database table.

A database that stores information that Metabase manages (as opposed to the application database that stores Metabase’s own management data).

Putting an iframe to a publicly-visible question in another web page.

Now deprecated in favor of Dashboard Subscriptions, pulses were questions that ran according to a specific schedule and would be sent out via email or Slack.

Learn about Query builder, the graphical interface for asking questions in Metabase.

In Metabase, a question is a query, its results, and its visualization.

RAG (retrieval-augmented generation) fetches relevant documents at question time and includes them in the prompt, so a language model answers from your content.

A copy of a database that receives a continuous stream of changes from the primary and serves read-only queries, used to spread load and to keep heavy reads away from production writes.

A group of related data with the same structure. A relational database stores each record as a row in a table.

A map that shows information for geographic regions (such as countries or states), e.g., by coloring each region to show a value.

A collection of tabular data, or the application that manages the storage and retrieval of tabular data.

The share of customers, users, or recurring revenue from a starting group that you still have at the end of a period.

Reverse ETL is the process of copying data out of your data warehouse and into the operational tools your teams work in every day, like a CRM or a support desk.

A join that keeps every row from the second (right-hand) table, regardless of whether that row matched any in the first (left-hand) table.

A single group of related data within a table.

A variation on a bar chart, where data is visualized with horizontal bars rather than vertical.

An XML-based standard for exchanging authentication and authorization data.

A scannable bar code used to uniquely identify some type of product.

A standardized and widely-used language for accessing and manipulating data in a relational database.

A named bit of SQL code that you can insert in a SQL query.

An authentication (auth) setup that lets people use one login to access independent apps.

A small database that ships with Metabase so you can start asking questions and creating dashboards before you connect to your database.

A sankey diagram shows how a quantity flows between stages or categories, with each link's width proportional to the size of the flow.

A Metabase question that has been saved and given a name. The results of saved questions can be used as the input to other questions.

A scatter plot places each data point at the intersection of two numeric values, revealing relationships, clusters, and outliers between the two variables.

The design or structure that defines the organization of a dataset, including its tables, columns, relationships, data types, and integrity constraints.

A filter that hides any rows in a table that don’t match some criteria.

A widget that can be added to a dashboard that applies a search filter to the values used in one or more cards.

A method for embedding an iframe displaying a question or dashboard that uses signed tokens for authentication.

A specific subset of a larger group of items, like a certain grouping of customers.

A semantic layer is a shared set of definitions for the metrics and dimensions in your data, sitting between the raw warehouse tables and the tools people use to ask questions.

The process of storing the state of an application or other object as a file so that it can be reloaded (or deserialized) later.

A measure of how someone interacted with a system, such as the number of pages they viewed in succession.

A random value generated by a server to identify a specific user for the duration of one interaction session.

Marking a record as no longer active or valid without actually removing it from the database, usually by setting a deleted_at timestamp or an is_deleted flag.

A bar chart in which several values are shown for each category by stacking bars sized according to each value.

A star schema is a way of organizing analytical data into one central fact table surrounded by dimension tables that describe it.

A stored procedure is a named block of SQL and procedural logic saved inside the database, run by calling it rather than sending the statements.

A subquery is a query nested inside another SQL query, whose result the outer query uses as a value, a list, or a table.

The result of an aggregation that gets saved in a database or data warehouse so that people can work with those precomputed metrics.

Data’s natural habitat. In a database, a table is a series of fields, with the values of those fields arranged in rows.

A card on a dashboard where you can enter Markdown text to give context for your dashboard. Text cards can also be used as dividers on a dashboard.

Text-to-SQL is the use of a language model to translate a natural-language question into a SQL query that runs against a database.

A time series is a sequence of points charted over time.

A database optimized for a large number of small operations rather than a few large transactions. OLTP usually relies on transactional databases.

An upsert is a database operation that inserts a row if it doesn't exist and updates it if it does, in a single statement.

Any value in a program or query that can change. In Metabase, variables in SQL queries get enclosed in double braces.

A vector database stores embeddings and answers similarity searches over them quickly, using approximate nearest neighbor indexes.

A vector embedding is a list of numbers a model produces to represent content, arranged so that similar content ends up close together in vector space.

Making a single machine more powerful — more CPU, more memory, faster disks — to increase performance, rather than adding more machines.

A query and its results that function like a virtual table in your database.

A waterfall chart is a visualization that shows how a series of positive and negative changes accumulate from a starting value to an ending total.

The business practice of applying branding to someone else’s product.

A window function is a SQL function that computes a value for each row based on a related set of rows, without collapsing those rows the way GROUP BY does.

A summary or visualization of a table or column generated automatically to give users a quick insight into their data.