What is a Record? | The Complete Guide to Evidence, Data, Documents, Databases, History and Institutional Memory

A record is a preserved representation of an event, observation, transaction, decision, state or communication that can be used later as evidence, memory or information. If you are asking “what is a record?”, the useful answer spans paper documents, database rows, photographs, laboratory notebooks, medical records, financial transactions, sensor logs, legal documents, historical archives and digital files. A record connects a later reader to something that happened earlier by preserving enough information to reconstruct what was observed, decided, received or done.

A complete explanation of records must distinguish a record from raw data, a document, a copy and the event itself. The record is not the original event. It is a representation created through a process. A temperature reading records one measured property under specified conditions; a signed contract records an agreement in a legally significant form; an audit log records selected system actions. Every record therefore has a creator, context, time, format, purpose and chain of custody that affect how strongly it can support later claims.

This guide explains records from first principles: evidence, documents, data records, fields, databases, logs, provenance, metadata, timestamps, authenticity, integrity, versioning, corrections, retention, archives, scientific records, medical records, financial records, legal records, historical sources, digital preservation, privacy, cybersecurity, audit trails and artificial intelligence. The goal is a durable mental model: ask what was recorded, who or what created the record, how the representation was produced, what context travelled with it, whether it changed and what conclusions the record can legitimately support.

What Is a Record? A Short Definition

A record is information fixed in a durable form because it provides evidence or memory of an activity, event, observation or state. The form can be paper, digital, photographic, audio, physical or database-based. What makes something a record is not merely that it contains information, but that it has been captured and preserved in relation to a real process or decision.

Why Records Matter

Records allow organisations and societies to act across time. A doctor can see earlier test results, a scientist can compare new measurements with old ones, an accountant can reconstruct transactions and a historian can examine documents created centuries ago. Without records, every system would depend on current human memory. Records create continuity by moving evidence from one moment to another.

Record Versus Event

An event is something that happens; a record is a preserved representation of that event. A payment occurs when value is transferred. The receipt and transaction entry are records of the payment. This distinction matters because records can be incomplete, duplicated or incorrect even when the underlying event happened exactly once. Evidence analysis should therefore ask both what happened and how the event was captured.

Record Versus Data

Data are recorded representations of observations or facts, while a record usually bundles data with enough identity and context to represent one event, entity or action. A database row may be called a record because fields belong together as one unit. Not all data are records in the institutional sense, and records can contain narrative text, signatures and relationships beyond simple field values.

Record Versus Document

A document is a bounded information object such as a report, letter or file. A document becomes a record when it is captured as evidence of an activity or decision within a recordkeeping context. A blank template is a document but not yet evidence of a completed transaction. Once completed, approved and preserved, the same form may become an official record.

Record Versus Copy

Digital systems can create perfect bit-for-bit copies, so “original” becomes more complicated than in paper environments. A trustworthy copy can function as an authoritative record if provenance, integrity and custody are preserved. What matters is whether the system can demonstrate that the content corresponds to the captured record and whether transformations such as migration have been controlled and documented.

Evidence

Records are often used as evidence because they preserve traces of past activity. Evidential strength depends on how the record was produced. An automatically generated timestamp may be strong evidence of when a server received a request, while a retrospective note written from memory has different limitations. A record should never be treated as infallible merely because it looks official.

Memory

Records create external memory. Human memory is selective and reconstructive, while records can preserve exact text, numbers or images. Records are not perfect memories either: they capture only what the recording system selected. A meeting minute may omit tone and informal discussion; a photograph captures one field of view. Good institutional memory combines records with knowledge of their scope.

Context

A record without context can become meaningless. A number such as 37.2 could be temperature, price or score. Context includes who created the record, why, when, where, under which procedure and how it relates to other records. Recordkeeping systems preserve context through metadata, file structures, links and classification so future users can interpret the evidence correctly.

Provenance

Provenance describes origin and history. For records, it identifies the creator, source system and transformations through which the record passed. Provenance lets users distinguish an original observation from a later transcription or summary. When records move between systems, provenance should travel with them so later users can reconstruct how the information reached its present form.

Metadata

Metadata are data about the record: title, creator, date, format, identifier, access conditions, version and relationships. Some metadata are visible to users, while others are generated automatically. Metadata make records searchable and interpretable. Losing metadata can leave the content intact while destroying much of its evidential value because future users cannot determine source, date or status.

Timestamps

A timestamp records a time associated with an event such as creation, receipt or modification. Timestamps depend on clocks, time zones and system configuration. “Created at” can mean different things in different applications. Reliable systems specify which event a timestamp represents and use synchronised time where ordering across systems matters.

Identifiers

Records need identifiers so they can be distinguished and referenced. File names are convenient but can change; stable database keys or persistent identifiers reduce ambiguity. An identifier should remain unique within its namespace. Good systems avoid using human-readable descriptions as the only identity because names, titles and labels can change over time.

Fields

A structured record contains fields representing properties such as date, amount, status or owner. Fields make records machine-processable but depend on definitions. A field called “date” might mean order date, delivery date or last update. Data dictionaries and schemas turn terse field names into explicit contracts so records can be interpreted consistently.

Database Records

In database language, a record often means one row or structured instance containing related fields. A customer record might contain an identifier, name and status. Relational databases distribute one real-world entity across several tables, so one row should not automatically be mistaken for a complete representation. Database design separates entities and relationships to reduce duplication and inconsistency.

Primary Keys

A primary key uniquely identifies a database row. Stable keys let systems update the correct record even when descriptive fields change. Natural identifiers such as email addresses can be tempting keys but may change or be reused. Surrogate keys separate technical identity from human attributes, though systems still need rules for resolving duplicate real-world entities.

Transactions

A transaction record captures a completed or attempted exchange, change or operation. Financial systems record amounts, parties, times and statuses; inventory systems record movements; databases record state changes. Transaction records become especially valuable when systems need to reconstruct who did what and whether an operation committed successfully after an interruption.

Logs

Logs are chronological records of events produced by software, machines or organisations. Application logs can record errors and requests; security logs can record access; laboratory logs can record equipment state. Logs are only as useful as their design. Excess logging creates noise and privacy risk, while insufficient logging makes failures impossible to reconstruct.

Audit Trails

An audit trail records significant actions and changes so later reviewers can reconstruct accountability. It may show who created, viewed, approved or modified a record and when. Audit trails should be protected from ordinary alteration; otherwise the evidence of change can be rewritten along with the changed content. Good auditability separates action from the record of the action.

Authenticity

Authenticity asks whether a record is what it claims to be. Evidence can include provenance, signatures, custody, system controls and consistency with related records. Authenticity is not the same as truth. An authentic letter can contain a false statement. Authentication establishes origin or identity; factual evaluation still asks whether the content accurately represents reality.

Integrity

Integrity means the record has not been altered in unauthorised or undetected ways. Access controls, hashes, signatures, immutable logs and controlled workflows can support integrity. Legitimate corrections do not violate integrity when they are authorised and documented. A trustworthy recordkeeping system preserves both the corrected state and evidence that a change occurred.

Completeness

A complete record contains the information required for its purpose, not necessarily every possible detail. A financial transaction may require amount, currency, parties and date; a scientific observation may require instrument settings and units. Completeness should be defined against use because unnecessary fields increase burden and privacy risk without improving evidence.

Accuracy

Accuracy concerns whether recorded content correctly represents the underlying event or state. A perfectly preserved record can be inaccurate if the original entry was wrong. Quality controls therefore operate both at capture and after storage. Validation, double entry, calibration and reconciliation help detect errors before they become durable institutional memory.

Reliability

A reliable record was created through a process that consistently captures information in a trustworthy way. Reliability depends on procedure, system design and human practice. A sensor that frequently fails silently produces less reliable records than one with calibration checks and error flags. Reliability belongs to the record-creation system as much as to the file itself.

Versioning

Versioning records successive states of an information object. Drafts, revisions and approved editions can coexist, so users need to know which version is current and which was effective at a given time. Version history supports accountability by showing how content changed. Without versioning, later users may cite a record that silently replaced an earlier decision.

Corrections

Records sometimes need correction. Good practice does not erase evidence that an earlier value existed when the history matters. Instead, systems can record the correction, reason, time and authority. This creates a transparent chain from error to repair. The goal is accurate current information without pretending the original mistake never occurred.

Deletion

Deleting a record removes it from active systems, but digital deletion can be complex because copies, backups and derived datasets may persist. Retention and deletion policies should define what deletion means operationally. Some records must be preserved for legal, scientific or historical reasons; others should be destroyed when their purpose ends to reduce privacy and security risk.

Retention

Retention specifies how long records should be kept. The correct period depends on operational need, law, accountability, historical value and risk. Keeping everything forever is not automatically safe or useful. Long retention increases storage, privacy exposure and discovery burden. Good schedules distinguish temporary working information from records with continuing value.

Disposition

Disposition is the controlled decision about what happens when a retention period ends: destruction, transfer to archives or another authorised outcome. Disposition should be documented so organisations can prove that records were neither destroyed prematurely nor retained without purpose. Legal holds can suspend normal disposition when records may be needed for investigation or litigation.

Historical Records

Historical records are surviving traces used to investigate the past. Letters, census data, maps, photographs, court files and newspapers provide different perspectives. A historical record reflects the interests and recording practices of its creator, so historians ask who produced it, for what purpose and what voices are absent. Evidence becomes stronger when different sources can be compared.

Primary Sources

A primary source is evidence created during or close to the period or event being studied, or by direct participants, depending on the discipline. Primary does not mean unbiased or automatically accurate. A diary can be a primary source while containing mistaken memories. Its value lies in its relationship to the event and the questions researchers ask of it.

Scientific Records

Science depends on records of observations, measurements, methods and analysis. Laboratory notebooks, raw instrument files, calibration records and code can all form part of the evidential chain. A final graph is insufficient if the route from measurement to figure cannot be reconstructed. Scientific recordkeeping protects reproducibility and makes correction possible.

Laboratory Notebooks

A laboratory notebook records what was done, observed and changed during research. Useful entries include dates, sample identifiers, settings, unexpected events and links to data files. A notebook is not merely a polished retrospective report. Its value comes from preserving contemporaneous context that may explain why a result differs from later expectations.

Sensor Records

Sensors create streams of measurements that become records only through calibration, timestamps, units and device metadata. A temperature series without knowing the sensor location or calibration can be misleading. Automated recording reduces some human transcription errors but introduces technical failures such as clock drift, missing packets and firmware changes that must be documented.

Medical Records

Medical records preserve clinical observations, diagnoses, test results, treatments and communications needed for continuity of care. They are high-stakes records because errors can affect future decisions. A record reflects what was documented, not every aspect of the patient. Clinical interpretation therefore combines recorded evidence with current examination and updated context.

Electronic Health Records

Electronic health records integrate clinical information across encounters and services. They improve retrieval and sharing but can propagate copied errors or outdated information if review is weak. Interoperability requires shared identifiers, vocabularies and data standards. Access controls and audit logs are essential because health records contain highly sensitive personal information.

Financial Records

Financial records document transactions, balances, obligations and supporting evidence. In accounting, reliable records allow transactions to be traced from source documents into ledgers and financial statements. Reconciliation checks whether different records agree. Strong controls reduce the risk that one erroneous or fraudulent entry flows unnoticed through the reporting system.

Receipts

A receipt is a record acknowledging that a transaction or transfer occurred. It may document amount, date, seller, buyer or payment method. A receipt is evidence of a transaction but may not prove every surrounding claim, such as the quality of an item or the identity of the person who ultimately used it. Scope matters when records are used as proof.

Legal Records

Legal systems rely on records such as contracts, filings, judgments and registers. Formal requirements can determine whether a document has legal effect. Authenticity, custody and admissibility may matter when a record becomes evidence. Legal recordkeeping therefore combines information management with rules about authority, signatures, notice and preservation.

Contracts as Records

A signed contract can record agreed terms and evidence commitments. The record is important because memory and later interpretations may diverge. Contract records usually need version control and identification of the final executed form. Drafts can be useful historical evidence, but they should not be mistaken for the binding version when the legal system distinguishes them.

Public Records

Public records are created or held by public bodies under legal and administrative frameworks. They support accountability, service delivery and institutional memory. Access rules vary because transparency must be balanced against privacy, security and other protected interests. The fact that government holds a record does not necessarily mean every part is open to unrestricted public access.

Administrative Registers

Registers are structured record systems tracking entities or events such as births, businesses, land or licences. They can become important statistical infrastructure, but administrative definitions reflect operational rules rather than research concepts. Analysts using register data must understand who enters the system, who is excluded and how definitions changed over time.

Educational Records

Educational records can include enrolment, attendance, assessments, accommodations and learning evidence. Scores are records of performance under specified conditions, not complete descriptions of a learner. Good educational recordkeeping preserves context and limits interpretation so one result does not become a permanent identity detached from task, support and time.

Photographs as Records

A photograph records light captured from a viewpoint at a moment, but it does not preserve the entire scene. Framing, exposure, lens, editing and selection influence interpretation. Metadata can preserve time and device information, though metadata can also be altered. Photographs are powerful evidence when provenance and context are considered rather than assuming visual appearance is self-explanatory.

Audio and Video Records

Audio and video can preserve speech, movement and timing unavailable in written summaries. They still represent selected perspectives: microphone placement, camera angle and editing affect what is captured. Compression and transcription may transform the record further. As synthetic media become easier to create, provenance and authentication become increasingly important.

Maps as Records

Maps can record boundaries, infrastructure and interpretations of space at a particular time. Historical maps are evidence of both geography and the mapping institution’s priorities. A map may omit informal settlements, disputed boundaries or features outside its purpose. Treating maps as records therefore requires understanding scale, projection, date and authority.

Natural Records

Nature can preserve records without human writing. Tree rings, ice cores, sediments, corals and fossils contain physical traces of past conditions. Scientists interpret these proxies using models linking measurable properties to environmental variables. A natural record is not a diary written by nature; it is material evidence whose signal must be separated from other processes that also affect it.

Geological Records

Rocks and sediments preserve a geological record of deposition, deformation, erosion and biological activity. The record is incomplete because later processes destroy or alter earlier material. Geologists reconstruct history by combining stratigraphy, fossils, radiometric dating, geochemistry and structural relationships. Missing layers are evidence too when erosion or non-deposition can be identified.

Fossil Records

The fossil record preserves traces of past life, but preservation is selective. Hard parts fossilise more readily than soft tissues, and some environments preserve organisms better than others. Evolutionary interpretation therefore accounts for sampling and geological bias. An incomplete record can still provide strong evidence when patterns are consistent across independent sites and methods.

Climate Records

Climate records include instrumental measurements and proxy evidence such as ice cores, tree rings and sediments. Instrument records provide direct measurements for recent periods, while proxies extend reconstruction further into the past. Each record type has resolution, calibration and uncertainty. Combining multiple sources helps distinguish local noise from broader climate signals.

Digital Records

Digital records are encoded as files, database entries, messages or system states. Their apparent immateriality hides physical dependence on storage media, software and formats. Digital records can be copied perfectly but also altered invisibly. Trust therefore depends on access controls, hashes, metadata, versioning and preservation systems that keep meaning accessible as technology changes.

File Formats

A file format defines how digital content is structured. A record stored in an obsolete proprietary format may survive physically while becoming unreadable. Preservation therefore considers not only bits but interpretability. Open specifications, migration and emulation are strategies for keeping records accessible when original software or hardware disappears.

Fixity

Fixity is evidence that a digital object has not changed unexpectedly. Cryptographic hashes can serve as fingerprints: if the file changes, the hash usually changes. Fixity checking detects alteration but does not prove that the original file was accurate or authentic. It protects continuity of a captured object after the point at which trust was established.

Digital Signatures

Digital signatures use cryptography to provide evidence that data were signed using a particular private key and have not changed since signing. They support authenticity and integrity but depend on key management and identity infrastructure. A technically valid signature proves a cryptographic relationship, not that every factual statement inside the document is true.

Chain of Custody

Chain of custody documents who controlled evidence or records from collection through transfer and analysis. It is especially important in forensic, legal and laboratory settings. Gaps do not automatically prove tampering, but they weaken confidence about whether the item presented later is the same one originally collected and whether handling could have affected it.

Access Control

Access control determines who may read, create, modify or delete records. Different permissions support confidentiality and integrity. Least-privilege design gives users only the access needed for their work. Sensitive records also require monitoring because legitimate credentials can be misused. Access should be revocable when roles change.

Privacy

Records about people can reveal identities, behaviour, health, finances and relationships. Privacy requires limits on collection, use, sharing and retention. Removing names does not automatically make records anonymous because combinations of attributes can enable re-identification. Privacy-aware recordkeeping minimises unnecessary data and preserves clear purpose.

Security

Record security protects confidentiality, integrity and availability against accidental loss and malicious attack. Encryption, backups, access controls, monitoring and incident response all contribute. Security priorities vary: an archive may emphasise long-term availability, while a medical record system must balance confidentiality with rapid access for authorised care.

Backups

Backups create recoverable copies after deletion, corruption or disaster. A backup is not a complete records-management strategy because it may contain outdated or unauthorised data and usually lacks the contextual controls of the live system. Recovery plans must define which version is restored and how changes after the backup are reconciled.

Disaster Recovery

Disaster recovery restores systems and records after major interruption. Critical records need redundancy, off-site protection and tested restoration procedures. A backup that has never been restored is only a promise. Regular recovery tests verify that data, software, credentials and dependencies can actually be reconstructed within required time limits.

Records and Artificial Intelligence

AI systems use records as training data, retrieval sources and operational context. Record quality affects model output. AI can also generate new records such as summaries, classifications and transcripts. Generated records should preserve provenance indicating that AI contributed, especially when later users might mistake generated text for contemporaneous human observation.

Automated Record Creation

Software can create records automatically from sensors, transactions and user activity. Automation increases volume and consistency but can also preserve errors at scale. System designers should document triggers, fields and failure modes so later users know whether the record reflects direct observation, inferred classification or a default generated by software.

AI Summaries and the Original Record

An AI summary is a derived representation, not a replacement for the underlying record when exact wording or evidence matters. Summaries can omit caveats or misstate relationships. High-stakes workflows should retain links to source records and allow users to inspect the evidence behind generated interpretations. Compression should not erase the route back to provenance.

Records and Search

Search systems index record content and metadata so users can retrieve relevant items. Ranking affects which records become visible, so retrieval is not neutral. Good search interfaces show enough metadata to distinguish current from obsolete versions and primary records from summaries. Discovery should lead users toward the authoritative record rather than merely the easiest text match.

Records and Knowledge

Records preserve evidence, but knowledge requires interpretation, comparison and understanding. An organisation can possess millions of records and still fail to learn if information cannot be found or if mistakes are repeatedly stored without analysis. Records become organisational memory when people can connect them to decisions, outcomes and later correction.

Records and Accountability

Accountability requires evidence of decisions and actions. Records show what authority was exercised, what information was available and which outcome followed. Poor recordkeeping can make later review impossible. Yet exhaustive surveillance is not the answer: accountable systems record the decisions and evidence necessary for legitimate oversight while respecting privacy and proportionality.

Records and Transparency

Transparency depends on records that are accessible, intelligible and sufficiently complete. Publishing large volumes of poorly organised files can create nominal openness without practical understanding. Useful transparency includes indexes, definitions, context and explanations of redactions or unavailable records. Access is meaningful when users can reconstruct the relevant decision or event.

Record Linkage

Record linkage connects records believed to refer to the same entity across datasets. Exact identifiers make linkage easier, while names and addresses require probabilistic matching. False matches can merge different people; missed matches can split one person into several records. Linkage therefore requires quality measures, privacy safeguards and uncertainty rather than assuming identity from superficial similarity.

Duplicate Records

Duplicates occur when the same entity or event is represented more than once. They can inflate counts, create contradictory states and confuse communication. Removing duplicates is not as simple as deleting identical rows because near-duplicates may contain complementary information. Entity-resolution rules should preserve provenance and document how a surviving record was chosen.

Missing Records

Absence of a record is not automatically evidence that an event did not happen. Records can be lost, never created, destroyed under policy or excluded from a system. Conversely, in a system known to capture events reliably, absence can be informative. Interpretation depends on understanding the recording process and its expected completeness.

False Records

A false record intentionally or accidentally represents events inaccurately. Fraudulent documents, fabricated data and mistaken entries are examples. Detection can use cross-checks, signatures, physical examination, statistical anomalies and comparison with independent sources. An apparently official format is not enough; trust comes from provenance and corroboration.

Record Drift

Records can become misleading even without being altered when the world changes around them. An address becomes outdated, a classification changes or a status no longer applies. Systems therefore distinguish historical truth from current truth. A past address may remain an accurate record of where someone lived then while being unsuitable as a current contact field.

Common Misconceptions About Records

“A record is the event itself.” It is a preserved representation of the event.
“Official records are always correct.” They can contain measurement, entry and interpretation errors.
“Digital copies are not originals.” In digital systems, authenticity depends on provenance and integrity more than physical uniqueness.
“More records always mean better evidence.” Quality, relevance and independence matter.
“Deleting a file deletes every copy.” Backups and derived systems may persist.
“A timestamp explains when everything happened.” It must be clear which event and clock the timestamp represents.
“A database row is the whole real-world entity.” It is one structured representation.
“A summary can replace the source.” Derived summaries may omit evidential detail.

A Practical Record Checklist

What event, state or decision does the record represent?
Who or what created it?
When and under which procedure?
What metadata and provenance are preserved?
Is the record complete enough for its intended use?
Has it been modified, corrected or migrated?
Which version is authoritative?
Who has access and who may alter it?
How long should it be retained?
What independent evidence can corroborate it?

How to Read a Record Critically

Begin with provenance before content. Identify the creator, purpose, date and recording process. Then inspect what the record actually shows rather than what you hope it proves. Look for missing context, later annotations, version information and relationships to other records. Finally, compare independent sources. Critical reading does not mean distrusting every record; it means matching confidence to how the record was produced and preserved.

Frequently Asked Questions About Records

What is a record in simple terms?

A record is preserved information about something that happened, was observed, decided or communicated, kept so it can be used later.

What is a database record?

It is a structured set of related field values, often represented as one row or entity instance within a database system.

What makes a record trustworthy?

Trust depends on provenance, reliable creation methods, authenticity, integrity, context and evidence that changes were controlled.

Is every document a record?

No. A document becomes a record when it is captured and preserved as evidence of an activity, transaction or decision within a relevant context.

Can records be corrected?

Yes, but good systems preserve an audit trail showing what changed, why, when and under whose authority when historical accountability matters.

Why are records kept?

They support operations, evidence, accountability, legal obligations, research, continuity and historical memory.

What is an archival record?

It is a record preserved for enduring administrative, legal, evidential, cultural or historical value rather than only short-term operational use.

Can AI-generated text be a record?

Yes, if an organisation captures it as evidence of an activity, but provenance should indicate how it was generated and link to underlying source information where relevant.

Authoritative Starting Points

Records management, archival science and digital-preservation standards provide formal frameworks for trustworthy records, retention, provenance and long-term access. Database disciplines define structured records and transaction integrity, while scientific and clinical fields specify domain requirements for evidence and documentation. Across these traditions, the same principle holds: a record is useful only when its content remains connected to the context that explains what it represents and how it came to exist.

What Is a Record? The Complete Idea

A record is durable evidence carried forward through time. It preserves a representation of an event, observation, decision, transaction or state so later people and systems can retrieve, compare and act on information that would otherwise disappear.

The strongest understanding keeps several distinctions clear: the record is not the event itself; preservation does not guarantee accuracy; authenticity is not truth; a digital copy can still be authoritative; metadata and provenance are part of meaning; a summary is not the source; and absence of a record is not always evidence of absence.

Once those distinctions are stable, records become easier to evaluate. Trace origin, understand capture, preserve context, control change, protect access, retain only as long as justified and keep the route back to source evidence. From laboratory notebooks and database rows to contracts and climate proxies, records are the infrastructure by which societies remember what happened well enough to learn, prove and continue.