Skip to content

Stockertim Dictionary · Cybersecurity · Technology

Data classification

Also called information classification, classification scheme, data labelling

In short

Sorting an organisation's information into a few levels, such as public, internal, confidential and restricted, so that the strongest protection goes to data whose exposure would cause most harm.

  • Advanced
  • Process
  • Checked

In detail

Not all information needs the same protection. A press release is meant to be public; customer account details, merger plans or encryption keys could cause serious harm if leaked. Data classification gives each type of information a level, with handling rules attached: who may see it, where it can be stored, whether it must be encrypted, how it may be shared and when it must be destroyed.

Schemes are usually kept simple so that staff can apply them. A common four-level model runs from public through internal and confidential to restricted or highly confidential. Each level is defined by the impact of disclosure, alteration or loss. Documents and systems carry labels, often applied automatically by software that spots card numbers, identity numbers or keywords, and these labels drive controls such as data loss prevention and access rules.

Classification underpins much of security and privacy work: it tells a firm where its most sensitive data lives, which systems need the strongest controls, which suppliers handle what, and what must be reported if a breach occurs. Its weaknesses are over-classification, which leads people to ignore labels, and out-of-date inventories as data is copied into new tools, including AI services.

Worked example

Bluewater Bank applies four classes to its data.

  1. Public: published interest rates and annual reports.
  2. Internal: staff directories and policies.
  3. Confidential: customer statements and loan files, encrypted and access-controlled.
  4. Restricted: unannounced results and encryption keys, limited to named people and logged.

Controls scale with harm, so effort is concentrated on the data that matters most.

In practice

Banks, insurers and governments run classification schemes; the UK government uses its own security classifications. Firms add rules on whether each class may be pasted into public AI tools or shared with suppliers. Data protection impact assessments often start from the classification.

Watch out

  • Classification in security means sensitivity levels, which differs from classification models in machine learning.
  • Labelling everything confidential makes the scheme meaningless.

Learn it properly

Official sources

Facts last checked on 4 November 2026.

Education, not advice: the Stockertim Dictionary explains what terms mean. It never recommends buying or selling anything.

English