Glossary of Data Terms

All of the basic terms you need to know when you’re working with data.

A

A/B Testing: Analysis in which randomized target customers receive variations of an item, e.g. an application interface element or web page, to measure the effect on a desired outcome, such as conversions. Also called Split Testing or Multivariate Testing.
Algorithm: An automated series of steps that executes a mechanizable function, e.g. a mathematical function in a computer program
Anomaly Detection: Analysis in which the analyst is alerted when events occur that fall outside of an established norm. Examples include Fraud Detection and Bot Detection.
Anonymization: Removal of information about people from a database that could be used to identify individuals.
Application Programming Interface (API): A set of commands provided for a software application that allows programmers to interact with it.
Artificial Intelligence (AI): Software that carries out complex tasks that humans would normally be required to perform, such as pattern recognition and decision-making.
Attribution: Analysis that seeks to discover what prior event caused the event being analyzed.
Average: See Mean.

B

Big Data: Data that, because of volume or complexity, is beyond the processing capacity of ordinary analytics tools.
Biometrics: Data on the body of a user collected by digital tools designed for the purposes of measuring health or athletic performance.
Bot: A computer program that operates autonomously, to carry out tasks for a user and/or to mimic the behavior of a person.
Bulk Data: Data that is uploaded to storage all at once, e.g. historical data. In contrast with Streaming Data.
Business Operations: Management of tangible and intangible business assets with the intention of deriving more value from them.

C

Clickstream Analytics: Analysis of user behavior based on their clicks on web pages.
Cloud: Computer services that are remotely hosted and accessed through the internet.
Clustering Analysis: The identification of groups of events linked by close proximity to each other, e.g. for the purpose of anomaly detection.
Click-Through Rate (CTR): The ratio of users who click on an item to the number of users who view that item.
Columnar: A database management system that stores data in columns rather than rows so as to speed query performance.
Connector: An API that enables a user to receive and/or send data to or from different computer programs.
Content Management System (CMS): A computer program that provides a simplified interface for creating and editing pages and/or records on a website.
Conversion: A user action defined as a desired outcome to be measured, e.g. a purchase of a product or a signup for a membership.
Cost Per Click (CPC): Advertising payment method that charges a certain amount every time a user clicks on an advertisement.
Cost Per Thousand (CPM): Advertising payment method that charges a certain amount for every thousand views of an advertisement.
Cross-Channel Analytics: Analysis of behavioral data from multiple channels (e.g. online, mobile, in-store) to provide a more complete picture of customer preferences for purposes of targeting and promotion.
Customer Data Platform (CDP): An analytics platform that centralizes First Party Data and enriches it with Second and/or Third Party Data for a more accurate and actionable view of customer behavior.

D

Database: A collection of information organized to allow efficient access, management, and retrieval.
Data Management Platform: An analytics platform that centralizes First Party Data for better tracking of advertising campaign performance.
Data Model: A conceptual or logical diagram of how data will need to flow in a computer program to fulfill its requirements.
Data Silo: A data source that is difficult to connect with other data because of dependency on Engineering resources or other constraints.
Data Warehouse: A central repository of business data for an organization.

F

Federated Database: A system in which multiple databases are linked together and can be interacted with as if they were a single database.
First Party Data: Information collected by a company about its own customers, from sources such as web and mobile usage tracking, CRMs and Business Analytics tools.
Funnel Analysis: Analysis of customer behavior in stages, typically starting with Awareness and ending with purchase or signup (Acquisition).

G

Geolocation Data: Device sensor data that tracks the physical location of a user.
Governance: Management of the quality, accessibility and security of data within an enterprise.
Growth Hacking: The use of rapid experimentation across marketing and product channels to drive growth.
Growth Marketing: See Growth Hacking.

I

Ingestion: The process of receiving data from a source and converting it into a format that can be accessed in a Data Warehouse.
Internet of Things (IoT): The web of communication between devices equipped with internet connectivity (Smart Devices).

J

Javascript Object Notation (JSON): A common data format consisting of strings of Key Value Pairs.

K

Key Value Pair: A common data format in which the data, or “Values”, are identified with text labels called “Keys”. Used for some NoSQL databases.
Keyword: A word used as Metadata to identify a web page or other content in a search.

L

Live Data: Data that is Connected with other relevant data, Current, and Easily Accessible to the people and processes in an organization that need to use it.
Lock-In: Inability to easily move data from one platform to another due to limitations imposed by a data storage provider, Business Application, or CRM.
Log: A text string that contains information about the state of a computer program or an event. Computer Data is composed of Logs.
Lookalike Modeling: Analytic process for targeting online advertising to website visitors with similar interests as a company’s existing customers.

M

Machine Learning: A category of Artificial Intelligence (AI) software in which the behavior of a program changes and improves based on exposure to new data.
Map/Reduce: A type of algorithm that breaks a data set apart so it can be processed on separate systems (Map) and then combines the data returned by those processes to create a report.
Mean: The sum of all the numbers in a set divided by the amount of numbers in the set. Also called the Average value of a set.
Median: The middle point of a number set, for which half the numbers in the set will be above and half below.
Metadata: Data that gives information about other data. On a web page, the Keywords for the page are metadata.
Migration: The process of moving a company’s existing data from one platform to another.
Model: See Data Model.
MongoDB: A popular open source NoSQL database.
Multi-Touch Attribution: A method of scoring customer touchpoints that tells you the likelihood that any given touchpoint, in any channel, contributed to a sale.
Multivariate Testing: See A/B Testing.
MySQL: A popular open source SQL database.

N

Normal Distribution: A bell-shaped graph that displays how items are ranked according to a randomly-distributed variable.
NoSQL: A general term that describes databases with structures and rules that differ from the row-and-column based SQL format.

O

Omnichannel Marketing: A marketing strategy that uses data from different sales channels to drive sales and optimize customer experience across other channels. An example is using data about a customer’s in-store shopping behavior to provide targeted email offer coupons.

P

Portability: Ability to move data, especially large amounts of different data formats, from one platform to another.
Predictive Analytics: A marketing function that uses machine learning to extrapolate customer traits to predict future purchasing behavior.
Presto: A SQL-based query language that allows very fast queries of large and distributed databases.

R

Recommendation Engine: Software that uses data derived from customers’ purchase habits to determine products to recommend, as on an e-commerce site.
Relational Database: A database organized in tables of rows and columns.
Return On Advertising Spend (ROAS): Advertising spend metric, expressed as a percentage of the revenue earned compared to the dollars spent on a campaign. Formula: Revenue / Cost x 100. If you spent $200 and made $300 on a campaign, your ROAS would be 300/200 * 100 = 150{5b4fe36bddf3b02e316fbf4108886dcb7ff194fea4a9c7a3e1d13675f87abf4b}.
Return On Investment (ROI): Spend metric, expressed as a percentage of profit on an investment. Formula: Revenue – Cost / Cost. If If you spent $200 and made $300 on a campaign, your ROI would be 100/200 * 100 = 50{5b4fe36bddf3b02e316fbf4108886dcb7ff194fea4a9c7a3e1d13675f87abf4b}.
Retargeting: Online advertising in which ads are shown to users based on their browsing history.

S

Scalability: Ability of a platform to automatically scale to meet user needs. If a platform stops working or performance degrades once usage goes beyond a certain threshold, it is not scalable.
Schema: In computer programming, the structure of a database.
Schema-Flexible: A type of data processing that allows data with different schema, including Schemaless data, to be inserted into the same database, aiding in connectivity.
Schemaless: Data that is not structured in rows and columns, e.g. NoSQL databases.
Segmentation: Division of a group of customers into subgroups based on shared characteristics in order to target them more precisely in marketing campaigns.
Second Party Data: Data on customers obtained from partner companies.
Software as a Service (SaaS): Business software that is hosted in the cloud, rather than being installed locally on a user’s computer.
Split Testing: See A/B Testing.
SQL: Standardized Query Language, a programming language that can be used to access data in a relational database. Also refers to the specific type of relational database system that can be queried using that language.
Standard Deviation: A measure of the amount of variation in a group.
Statistical Significance: A strong enough correlation between two or more variables to be confident the correlation is not due to chance.
Streaming Data: Data that consists of logs generated dynamically by software activity. In contrast with Bulk Data.

T

Third Party Data: Aggregated Customer Data from an external vendor used to enrich a company’s First Party Data for better Segmentation in a Customer Data Platform.
Treasure ML: A library for implementing Machine Learning in Treasure Data.
Treasure Workflow: A Treasure Data feature that allows complex workflows to be constructed, scheduled and reused.