Understand digital work

120 essential definitions for better decisions.

A plain-language glossary for web, design, marketing, data and artificial-intelligence projects. Each definition adds a practical use or caution to support clearer briefs and conversations.

How to use it

Shared language creates clearer projects.

Use these definitions to prepare a brief, review a proposal or ask a more precise question.

Direct definitions
A concise answer followed by practical context.
Project relevance
Concepts connected to delivery, visibility and governance.
Independent reading
Open one term or browse the full collection.
Editorial approach
General explanations without automatic tool recommendations.

Results: 120

A to W

The digital glossary.

Terms are ordered alphabetically and explained for project decisions rather than as isolated technical theory.

SEO

301 redirect

A 301 redirect tells clients and search systems that a URL has permanently moved to a new address.

During a migration, it sends people and crawlers to the corresponding page rather than a dead end.

↑ Back to top
Technology

404 error

A 404 response means the requested resource cannot be found at that address.

A useful 404 page supports onward navigation; important old URLs should redirect when a relevant replacement exists.

↑ Back to top
Technology

503 error

A 503 response signals that a service is temporarily unavailable.

It suits a temporary interruption and may include a retry period, unlike a permanently removed page.

↑ Back to top
Project

Acceptance criterion

An acceptance criterion describes the observable result needed to approve a feature or deliverable.

Write it before implementation with a scenario, a testable condition and a named decision maker.

↑ Back to top
Quality

Acceptance testing

Acceptance testing compares a site’s behaviour with expected outcomes in concrete scenarios.

Describe, prioritise, fix and retest each discrepancy before approving a release.

↑ Back to top
Visibility

AEO

Answer Engine Optimisation describes work that makes an answer clear, reliable and usable by answer engines.

It does not replace SEO: the page must still be helpful, accessible, indexable and supported by evidence.

↑ Back to top
Artificial intelligence

AI system evaluation

AI evaluation compares a system’s outputs with criteria defined for a particular use.

Measure errors, robustness and the cost of human review before scaling a prototype.

↑ Back to top
Images

Alternative text

Alternative text describes the purpose or essential information of an image for people who cannot see it.

Its wording depends on context: a decorative image may have an empty alt attribute, while an informative chart needs an explanation.

↑ Back to top
Data

Analytical unit

The analytical unit is the object described by observations and results, such as a person, enterprise, establishment or territory.

Counting establishments and counting enterprises answer different questions; state the chosen level.

Apply this term to a decision →

↑ Back to top
Personal data

Anonymisation

Anonymisation aims to prevent a person from being identified from a dataset over time.

Assess the risk of joining it with other sources before calling a dataset anonymous.

↑ Back to top
Technology

API

An API is an interface that lets software exchange data or trigger actions through documented rules.

It can connect a website to a payment service, CRM or trusted data source.

↑ Back to top
Data

API pagination

Pagination splits a result collection across several responses instead of transferring everything at once.

An import must traverse the required pages and account for changes to avoid missing or duplicate records.

↑ Back to top
Data

API quota

An API quota limits calls or consumption over a period defined by the provider.

Check the limit, scope, response on exceedance and retry strategy before integrating.

↑ Back to top
Measurement

Attribution

Attribution is the method used to connect a conversion to earlier marketing touchpoints.

No model fully explains a complex decision, so document assumptions and limitations.

↑ Back to top
Operations

Automation

Automation delegates repeatable tasks to a system that follows defined triggers and rules.

A maintainable automation has an owner, monitoring, documented exceptions and a manual fallback.

↑ Back to top
Images

AVIF

AVIF is a compressed image format for the web that can be effective at reducing visual file sizes.

Provide a fallback format where the delivery context or tools require one.

↑ Back to top
Performance

Cache invalidation

Cache invalidation makes a stored version obsolete so an updated resource can be retrieved.

Versioned files support renewal; test after publication that visitors receive the intended version.

↑ Back to top
SEO

Canonical URL

A canonical URL is the preferred version indicated to search systems when several addresses contain the same or very similar content.

It supports consistent redirects and internal links, but cannot repair a confused architecture by itself.

↑ Back to top
Performance

CDN

A content delivery network serves files from distributed locations closer to visitors.

It can reduce latency for images and scripts but does not replace file optimisation.

↑ Back to top
Performance

CLS

Cumulative Layout Shift measures unexpected movement of visible elements while a page is being used.

Reserving space for images and late-loading components helps prevent jumps.

↑ Back to top
Web

CMS

A content management system helps teams create, organise and publish pages without editing every page in code.

Selection depends on editorial workflow, permissions, integrations, maintenance and customisation needs.

↑ Back to top
Measurement

Cohort

A cohort is a group defined by a common criterion and studied over a specified period.

For employment figures, retain period, size, response rate and definition of measured employment.

Apply this term to a decision →

↑ Back to top
Content

Content audit

A content audit inventories pages and assesses their usefulness, accuracy and role in user journeys.

It helps teams keep, improve, merge or remove material before a redesign.

↑ Back to top
Trust

Content provenance

Content provenance explains who created or published information and which sources support it.

Visible authorship, dates, methods and corrections help readers assess reliability.

↑ Back to top
SEO

Contextual linking

Contextual linking connects one page to another where a reader needs a related explanation.

Descriptive link text makes the destination clear and helps important pages be discovered.

↑ Back to top
Marketing

Conversion

A conversion is a useful action completed by a visitor, such as an enquiry, sign-up, download or purchase.

Its value depends on a clearly defined outcome, not on the number alone.

↑ Back to top
Performance

Core Web Vitals

Core Web Vitals assess loading, responsiveness and visual stability through LCP, INP and CLS.

They are useful experience indicators but do not describe the complete quality of a service.

↑ Back to top
Operations

CRM

A CRM centralises information and interactions concerning prospects, customers or partners.

It supports follow-up and shared history when access, quality and retention are governed.

↑ Back to top
Data

Data catalogue

A data catalogue lists available datasets with their owner, definition and access conditions.

Add update frequency, licence and known limitations before inviting reuse by other teams.

↑ Back to top
Data

Data dictionary

A data dictionary describes the meaning, format, units and expected values of dataset fields.

A column called date may refer to publication or observation; inspect its definition before comparing.

Apply this term to a decision →

↑ Back to top
Data

Data governance

Data governance defines who may create, access, change, secure and delete organisational data.

It also sets ownership, quality rules, retention and trusted sources.

↑ Back to top
Data

Data join

A join relates records from different datasets using keys and matching rules.

Check unique keys, one-to-many relationships and unmatched records to avoid multiplying or losing observations.

Apply this term to a decision →

↑ Back to top
Data

Data lineage

Data lineage records where data came from, how it was transformed and where it is used.

It helps explain a figure, correct a source and assess the impact of a processing change.

↑ Back to top
Data

Data quality

Data quality describes whether information is fit for a specific use through accuracy, completeness, consistency and freshness.

Measure defects that change decisions; a large dataset can still be unsuitable for a critical task.

↑ Back to top
Data

Data schema

A data schema describes the fields, types, relationships and constraints in a dataset.

Share it between producers and users to reduce integration and migration errors.

↑ Back to top
Project

Decision log

A decision log records significant project choices with their date, context and owner.

It prevents the same discussion being reopened and makes the trade-offs understandable later.

↑ Back to top
Design

Design system

A design system combines shared principles, interface components and usage rules for consistent experiences.

It improves collaboration but does not replace user research or context-specific design decisions.

↑ Back to top
Quality

Digital accessibility

Digital accessibility means designing content and services that more people, including disabled people, can use.

It covers structure, keyboard use, contrast, forms, media and assistive-technology compatibility.

↑ Back to top
Efficiency

Digital ecodesign

Digital ecodesign integrates environmental-impact reduction into the design and operation of a service.

It examines usefulness, uses, resources and dependencies; page weight does not capture the whole approach.

↑ Back to top
SEO

Duplicate content

Duplicate content is identical or very similar material accessible through several URLs.

Identify the version useful to readers, align links and canonicals, and consolidate variants without distinct value.

↑ Back to top
Editorial quality

E-E-A-T

E-E-A-T groups experience, expertise, authoritativeness and trust concepts used when assessing content quality.

It is not a public score or meta tag; show responsibility, sources, evidence, limits and relevant first-hand experience.

↑ Back to top
Content

Editorial review

An editorial review checks whether a page still meets its purpose and whether facts, links and visuals remain accurate.

An updated date should reflect a real review rather than a timestamp change alone.

↑ Back to top
Artificial intelligence

Evaluation set

An evaluation set contains representative cases used to measure a system’s behaviour.

Include difficult examples, expected errors and real use cases rather than only a polished demonstration.

↑ Back to top
Images

EXIF

EXIF is metadata embedded in some image files, including capture settings and descriptions.

Remove sensitive fields before publication and do not use EXIF as a substitute for page alt text.

↑ Back to top
Budget

Exit cost

Exit cost covers spending required to leave a solution and resume work elsewhere.

Include export, migration, handover and internal effort; ask whether exported files are actually reusable.

Apply this term to a decision →

↑ Back to top
Search

Faceted navigation

Faceted navigation combines multiple criteria to narrow a list of results.

Empty states, usability and generated URLs need review to avoid confusing journeys.

↑ Back to top
Measurement

Field data

Field data describes the experience of real visitors on their own devices and networks.

It complements laboratory tests and requires a clear collection method and respect for privacy.

↑ Back to top
Data

First-party data

First-party data is collected directly by an organisation through its own relationships and services.

It still needs a defined purpose, proportionate collection, protection and respect for people’s rights.

↑ Back to top
Personal data

GDPR

The GDPR governs personal-data processing in the European Union and strengthens people’s rights.

Organisations need clear purposes, proportionate collection, secure access and transparent information.

↑ Back to top
Artificial intelligence

Generative AI

Generative AI produces text, images, audio, code or other material from instructions and learned patterns.

Outputs require risk-based checking because they can be incorrect, incomplete or unsuitable.

↑ Back to top
Visibility

GEO

Generative Engine Optimisation aims to make content understandable, reliable and usable in generative answer experiences.

Clear entities, direct answers, original evidence, sources and crawlable technical foundations support this work.

↑ Back to top
Data

Geocoding

Geocoding associates a geographic location with a description of a place, such as an address.

An ambiguous match requires review; a coordinate establishes neither identity nor someone’s presence.

Apply this term to a decision →

↑ Back to top
International SEO

Hreflang

Hreflang connects equivalent language or regional versions so search engines can present the appropriate page.

Each page should reference itself and its equivalents with accessible, canonical and reciprocal URLs.

↑ Back to top
Performance

HTTP cache

HTTP caching allows a previously fetched response to be reused for a defined time.

Appropriate cache rules speed up stable assets while allowing changed content to refresh.

↑ Back to top
Technical

Idempotence

An idempotent operation has the same intended effect when repeated with the same parameters.

This property supports safe retries; repeated payment or message requests should specifically avoid duplicates.

↑ Back to top
Images

Image compression

Image compression reduces the bytes needed to store and deliver a visual.

Compare the result at the real display size; a small file is useful only when important details remain legible.

↑ Back to top
Rights

Image licence

An image licence states which uses the rights holder permits for a visual.

Check alteration, distribution and credit conditions before using a photo in a guide, card or social preview.

↑ Back to top
Images

Image metadata

Image metadata is information embedded in a file, such as its creator, description or camera settings.

It helps manage an asset but can be removed by processing; visible context and alternative text remain essential.

↑ Back to top
Images

Image sitemap

An image sitemap tells search engines which visuals are associated with public pages.

It helps discovery but does not guarantee an image will appear in search results.

↑ Back to top
Indexing

IndexNow

IndexNow is a protocol for notifying participating search engines when a URL is added, changed, moved or deleted.

It complements an XML sitemap, which remains the durable inventory of discoverable preferred URLs.

↑ Back to top
Experience

Information architecture

Information architecture organises content, labels and navigation paths within a service.

It should help people find answers without knowing the organisation’s internal vocabulary.

↑ Back to top
Performance

INP

Interaction to Next Paint assesses a page’s overall responsiveness during user interactions.

Slow responses may call for less JavaScript work or better scheduling of tasks.

↑ Back to top
SEO

Internal linking

Internal linking connects pages on the same site so people and crawlers can understand their relationships.

Descriptive anchors and contextual links make important material easier to discover.

↑ Back to top
Architecture

Interoperability

Interoperability is the ability of separate systems to exchange data and use it correctly.

Check formats, identifiers, update rules and error handling, not merely whether an API exists.

↑ Back to top
Images

IPTC

IPTC metadata can record a visual creator, rights, description and licence information.

It helps rights management when the file format and publishing pipeline retain the metadata.

↑ Back to top
Operations

Isolated restoration

Isolated restoration recovers a copy in a separate environment to verify its data and operation.

Disable outgoing actions such as email or payment so the test does not trigger real operations.

↑ Back to top
Structured data

JSON-LD

JSON-LD adds machine-readable structured data without mixing it into the visible HTML content.

Declared entities and properties must match information people can actually see on the page.

↑ Back to top
Measurement

KPI

A KPI is an indicator selected to track a specific objective and support a decision.

A useful KPI has a definition, owner, context and possible action.

↑ Back to top
Performance

Lazy loading

Lazy loading delays retrieval of a media item farther down the page until it becomes useful.

Avoid delaying the main image visible on arrival because that can slow its appearance.

↑ Back to top
Performance

LCP

Largest Contentful Paint measures how long it takes to display the largest visible element in the initial viewport.

The main image, server response and blocking resources can influence this measure.

↑ Back to top
Artificial intelligence

LLM

A large language model is trained to process and generate language from large collections of data.

It can summarise or classify material but does not guarantee factual accuracy or human understanding.

↑ Back to top
Decision

Mandatory criterion

A mandatory criterion excludes an option when an essential requirement is absent, regardless of its scores elsewhere.

Assess it separately from a weighted average so a major incompatibility is not concealed.

Apply this term to a decision →

↑ Back to top
Management

Measurement plan

A measurement plan connects outcomes, tracked events, indicator definitions and decisions.

Assign an owner to each measure and check that it works before using it in a dashboard.

↑ Back to top
Images

Mobile crop

A mobile crop adapts an image composition to a narrow screen while keeping its essential subject.

Check the face, object or central information at several widths before publishing a variant.

↑ Back to top
Product

MVP

A minimum viable product is a deliberately limited but useful version built to test an assumption with real users.

It is a learning scope, not an excuse for unreliable quality.

↑ Back to top
Visibility

Named entity

A named entity is an identifiable person, organisation, place or concept in content.

Consistent naming, precise descriptions and links to official sources reduce ambiguity.

↑ Back to top
Tools

No-code

No-code uses visual interfaces and configurable components to build sites, applications or automations.

It can accelerate delivery but still needs decisions about data, security, ownership and maintenance.

↑ Back to top
Operations

Observability

Observability helps a team understand a service through its measurements, logs and traces.

Choose signals tied to critical journeys and assign someone to investigate alerts.

↑ Back to top
Rights

Open Licence 2.0

France’s Open Licence 2.0 permits reuse of public information, including copying, adaptation and commercial exploitation, under its terms.

Check that it applies to the dataset and retain producer attribution and the information’s update date.

Apply this term to a decision →

↑ Back to top
Performance

Performance budget

A performance budget sets limits for transferred weight, requests or time on a page template.

It helps govern new images and third-party services, using representative mobile checks.

↑ Back to top
Editorial quality

Primary source

A primary source directly publishes data, a rule, research or documentation for which it is responsible.

Trace claims to the source where possible and check its date, scope and limitations.

↑ Back to top
Project

Project dependency

A dependency is an input or decision required before a task can proceed or be accepted.

Content, access, legal decisions and third-party APIs belong in the schedule with named owners.

↑ Back to top
Personal data

Pseudonymisation

Pseudonymisation replaces direct identifiers with separate references without necessarily preventing re-identification.

Protect the mapping table and continue to handle the information as personal data.

↑ Back to top
Artificial intelligence

RAG

Retrieval-augmented generation combines a generative model with information retrieved from selected sources.

Quality depends on retrieval, document freshness, access rights, citations and answer verification.

↑ Back to top
Web

Responsive design

Responsive design adapts a layout to different screen sizes and device capabilities.

The interface should remain readable, operable and complete across mobile, tablet and desktop.

↑ Back to top
Images

Responsive image

A responsive image offers a suitable size or crop for the screen and layout.

Srcset or a picture element can supply variants while an img element provides a fallback URL.

↑ Back to top
Accessibility

RGAA

RGAA is the French reference framework used to assess digital accessibility and organise compliance work.

It provides test criteria and methods within the applicable French legal context.

↑ Back to top
Efficiency

RGESN

RGESN is the French reference framework for ecodesign of digital services.

It supports examination of service choices; citing it does not establish that its criteria have been assessed.

↑ Back to top
Project

Risk analysis

Risk analysis identifies events that could prevent an outcome and considers their consequences.

For each priority risk, record a warning signal, a mitigation and the person monitoring it.

↑ Back to top
Training

RNCP

France’s National Register of Professional Certifications lists qualifications that lead to an occupation.

Inspect the exact record for validity, level, skills and authorised organisations before comparing training.

Apply this term to a decision →

↑ Back to top
Crawling

Robots.txt

Robots.txt tells crawlers which areas of a site they may or may not request.

It does not secure private information or replace a noindex directive on an accessible page.

↑ Back to top
Launch

Rollback

A rollback restores a previous version when a release causes a major problem.

The procedure should identify backups, owners and data that may have changed since deployment.

↑ Back to top
Continuity

RPO

RPO expresses the maximum amount of data an organisation is prepared to lose after an incident.

It informs backup frequency and should be tested through a realistic recovery scenario.

↑ Back to top
Continuity

RTO

RTO expresses the target time for restoring a service after an interruption.

Compare that target with actual restoration capacity, dependencies and staff availability.

↑ Back to top
Performance

RUM

Real User Monitoring measures the experience of authorised real visits to a website.

It complements laboratory tests by reflecting varied devices, networks, pages and interactions.

↑ Back to top
Content

Search intent

Search intent is the need a person is trying to meet behind a query.

A useful page addresses that need with suitable depth, examples and a clear next step.

↑ Back to top
Data

Selection bias

Selection bias occurs when observed data represents some situations or populations more than others.

An average may hide missing audiences; check coverage before generalising a result.

↑ Back to top
Visibility

SEO

Search engine optimisation helps relevant pages be crawled, understood and found in search.

It combines useful content, architecture, technical quality, internal links and credible evidence.

↑ Back to top
Operations

SLA

A service level agreement defines support or availability commitments.

It should state how results are measured, exclusions, response times and responsibilities.

↑ Back to top
Operations

SLO

An SLO is a measurable service-level objective for a defined period and user population.

Specify what counts as success, the measurement window and the action taken when the objective is missed.

↑ Back to top
Quality

Staging environment

A staging environment lets a team review a site version before visitors see it.

It should reproduce important journeys without exposing unnecessary real data or being indexed accidentally.

↑ Back to top
SEO

Structured data

Structured data explicitly describes entities and relationships with a shared vocabulary such as Schema.org.

It can improve understanding but does not guarantee a rich result or ranking.

↑ Back to top
Experience

SXO

Search experience optimisation connects discoverability with the quality of the journey after a visit begins.

It aligns search intent with clarity, trust, speed and task completion.

↑ Back to top
Operations

Synchronisation

Synchronisation reconciles data copies to propagate changes across locations.

It can also propagate deletion; check recoverable versions before treating it as a backup.

↑ Back to top
Content

Taxonomy

A taxonomy is a coherent set of categories and terms used to classify content or resources.

Stable classification rules improve filters, internal links and editorial maintenance.

↑ Back to top
Maintenance

Technical debt

Technical debt is the future cost created by implementation choices that make later changes harder.

Manage it by documenting trade-offs and reserving time for valuable corrections.

↑ Back to top
Budget

Total cost of ownership

Total cost of ownership covers spending needed to acquire, operate, maintain and exit a solution over a defined period.

State the duration and assumptions; compare licences, integration, support, internal effort and data export consistently.

↑ Back to top
Design

UI

User interface refers to the visual and interactive elements through which someone uses a digital product.

Controls, typography, colour and states should form a coherent accessible language.

↑ Back to top
Operations

Uptime

Uptime is the share of a measured period during which a service responds correctly to requests.

Its meaning depends on the period and criteria; a page that responds but cannot be used is still a problem.

↑ Back to top
Experience

Usability test

A usability test observes someone completing a representative task with a product or prototype.

Record what they understand, where they hesitate and whether they finish unaided, then fix recurring obstacles.

↑ Back to top
Experience

User journey

A user journey describes the steps a person takes to complete a goal in a service.

It reveals questions, obstacles and information needed at each point.

↑ Back to top
Experience

User research

User research investigates the needs, constraints and behaviour of people affected by a service.

Combine open questions with task observation; stated preference is not the same as successful use.

↑ Back to top
Design

UX

User experience describes the overall quality a person experiences before, during and after using a service.

It depends on usefulness, ease, accessibility, trust and real context.

↑ Back to top
Accessibility

WCAG

WCAG are W3C recommendations for making web content more perceivable, operable, understandable and robust.

They underpin many accessibility standards and should be combined with human testing and governance.

↑ Back to top
SEO

Web crawling

Crawling is the process through which a robot discovers and requests URLs, links and resources to understand a site.

Clear HTML links, current sitemaps and reliable server responses support discovery.

↑ Back to top
Technology

Webhook

A webhook sends information from one service to another when a defined event occurs.

It can report a payment or form submission so another process begins automatically.

↑ Back to top
Images

WebP

WebP is a web image format that can reduce the size of a photograph or illustration at comparable visual quality.

The outcome depends on the image and compression level, so inspect the export.

↑ Back to top
Indexing

XML sitemap

An XML sitemap lists the canonical URLs a site wants search engines to discover.

It helps signal new or changed pages but does not guarantee crawling or indexing.

↑ Back to top
Apply the language

Turn definitions into decisions.

Explore the expertise families, then use the practical guides to prepare a clearer request.

Move from a definition to a practical decision

For technical definitions, inspect the reference documentation as well as the project context. MDN · Idempotence · W3C WAI · Images · RGESN.