CPCU 550 key terms

The set of policies, roles, and decision rights that determine who may collect, change, share, and retire an insurer's data. It assigns accountability for data quality and defines how disputes over data use are resolved across underwriting, claims, and analytics teams.

A day-to-day operational role in which a named individual is accountable for the accuracy, completeness, and appropriate use of a defined data set. Stewards resolve quality issues, approve changes, and act as the bridge between business units and technical teams.

The degree to which data is accurate, complete, timely, consistent, and fit for its intended use. Poor quality shows up as mispriced policies, failed regulatory reports, or claims paid to the wrong party, so insurers measure it with defined dimensions and remediation targets.

Structured information that describes other data, such as its source, meaning, format, owner, and refresh schedule. Business metadata explains what a field means to underwriters; technical metadata tells engineers how and where the value is stored.

A documented trace of where a data element originated, how it moved and transformed across systems, and where it ended up. Insurers use it to investigate reporting errors, support audits, and assess the impact of a system change before it is made.

The record of the origin and history of a data element, including who created it, when, and through what process. It supports defensibility when a regulator or litigant challenges how a claim, rating, or reserve figure was produced.

A reference catalog listing each data element, its definition, allowed values, format, and owning system. It prevents the same label, such as 'incurred loss', from meaning different things in different departments and supports consistent reporting.

A discipline for creating and maintaining a single, authoritative version of core entities such as customers, policies, agents, and locations. It reduces duplicate records and conflicting attributes that arise when each business system keeps its own copy.

The process of detecting and correcting or removing inaccurate, incomplete, duplicated, or malformed records. Insurers typically run it during migration and periodically thereafter, documenting the rules applied so results are repeatable and auditable.

Converting data from one format, structure, or scale into another so it can be used by a downstream system or analysis. Examples include standardizing date formats, unit conversions, and translating coded values into readable labels during an extract.

Extract, transform, and load: a pipeline pattern that pulls data from source systems, applies cleaning and business rules, and writes the result to a target store. Insurers use it to populate warehouses and reporting marts on a schedule.

The automated sequence of steps that moves data from sources to a destination, including extraction, validation, transformation, and loading. Pipelines run on schedules or events and are monitored for failures that could leave reports stale.

A centralized repository that consolidates data from multiple operational systems, structured for query and analysis rather than transaction processing. It typically holds historical snapshots so trends and loss development can be analyzed over time.

A focused subset of a data warehouse built for one business area, such as claims or personal lines pricing. Marts simplify access and improve performance for a defined user group while inheriting definitions from the enterprise warehouse.

A storage repository that holds large volumes of raw data in native formats until a use is identified. It offers flexibility for unstructured sources such as images and text, but without governance it can become an unreliable, poorly documented collection.

Data organized into predefined fields, rows, and columns, such as policy records in a relational database. It is easy to query and validate, and most financial and regulatory reporting in insurance relies on it.

Data with no fixed field structure, such as claim photos, recorded calls, email text, and scanned documents. Insurers increasingly analyze it with text and image tools, which requires different storage, search, and quality controls than tabular data.

A data store that organizes information into tables linked by defined keys, queried with a standard language such as SQL. Its constraints help enforce consistency, which is why policy administration and claims systems commonly rely on it.

The assurance that data remains accurate, consistent, and unaltered except through authorized, recorded changes. It covers both physical protection from corruption or loss and logical rules that prevent invalid combinations of values.

A chronological record of who accessed or changed a record, what changed, and when. Insurers rely on it to demonstrate controls to regulators, investigate internal fraud, and reconstruct the state of a claim or policy at a past date.

The rules and mechanisms that determine which users or systems may view or modify specific data. Common models include role-based access, where permissions follow job function, and attribute-based access, where context such as location or purpose is evaluated.

A security principle under which each user, service, or application receives only the minimum access needed to perform its function. Applied to insurance data, it limits exposure of sensitive claim and medical information to staff who do not need it.

The safeguards that protect data from unauthorized access, disclosure, alteration, or destruction. Controls include encryption, access restrictions, monitoring, vendor oversight, and incident response, and they apply wherever data is stored or transmitted.

Encoding data so that only parties holding the correct key can read it. Insurers use it for data at rest in databases and at motion across networks; breach-notification laws often treat strongly encrypted lost data more leniently.

An incident in which protected data is accessed, disclosed, or lost without authorization. Insurers face notification duties, regulatory scrutiny, remediation costs, and reputational harm, and many also insure others against such events.

Information that can identify a specific individual, alone or combined with other data, such as name, address, Social Security number, or account details. Insurers hold large volumes of it and must handle it under privacy and security laws.

The rules governing how personal information may be collected, used, shared, and retained, and the rights individuals hold over it. Insurers must align marketing, underwriting, and claims practices with applicable privacy statutes and their own notices.

A schedule specifying how long each category of records must be kept and when it must be destroyed. It balances legal and regulatory minimums against privacy limits and litigation holds that suspend destruction.

Data sets too large, fast-moving, or varied for conventional tools to process comfortably. Insurers apply it to telematics, sensor feeds, and web activity, often described by the dimensions of volume, velocity, variety, and veracity.

The use of historical data and statistical or machine-learning models to estimate the likelihood of future outcomes. Insurers apply it to pricing, fraud detection, claims triage, and retention, and must document it for regulatory review.

A class of methods in which algorithms learn patterns from data and improve with experience rather than through explicit rule coding. In insurance it supports risk scoring, document classification, and claims severity prediction.

A defined sequence of steps or rules used to compute a result from inputs. Pricing algorithms, claims assignment routines, and fraud scores are all examples, and their logic may need to be explained to regulators and customers.

The potential for loss or harm from decisions based on a model that is wrong, misapplied, or built on unrepresentative data. Insurers manage it with validation, documentation, monitoring, and defined ownership.

Systems that perform tasks associated with human intelligence, such as language understanding, image recognition, and decision support. Insurers use it for document intake, customer service, and claims triage, subject to governance and disclosure expectations.

A field of computing concerned with enabling machines to read, interpret, and generate human language. Insurers apply it to classify claim narratives, extract terms from policies, and route incoming correspondence.

Software that mimics user actions to perform repetitive, rule-based tasks across systems, such as copying data between screens. It is fast to deploy but brittle when interfaces change and is best suited to stable, high-volume work.

An older application still in production that is costly to maintain, poorly documented, or hard to integrate. Insurers often keep core policy and claims platforms for decades, so data extraction and interface work around them is a common project.

The primary platform that runs essential insurance transactions, such as policy administration, billing, or claims. Because it holds authoritative records, changes to it require careful testing, migration planning, and coordination with reporting.

The accumulated cost of shortcuts in system design or data structure that make future changes harder. Examples include hard-coded rules and undocumented interfaces, which slow modernization and raise the risk of errors during updates.

Delivery of computing resources such as storage, processing, and software over a network on demand. Insurers use it for scalability and analytics, while addressing data residency, vendor oversight, and exit-planning requirements.

A delivery model in which a vendor hosts an application and customers access it over the internet, usually by subscription. Insurers must assess the provider's security, data ownership terms, availability commitments, and integration with core systems.

The process of selecting, contracting with, monitoring, and reviewing third parties that provide services or technology. For data-related vendors, it includes due diligence on security, defined service levels, and rights to data on termination.