Module 1 of 26

What data is

Raw facts, symbols, and signals: what data actually is before it becomes anything useful, and why the distinction between data, information, and knowledge matters for everyone who works with it.

By the end of this module you will be able to:

  • Distinguish data from information and knowledge using concrete examples
  • Classify a dataset as structured, unstructured, or semi-structured
  • Explain the scale of global data growth using current estimates

How observation becomes information that survives the journey

Raw observations only become useful when representation, record, and decision use travel together with their metadata.

Data only becomes useful when meaning survives the journey Two regions. Top row: four cards Observation, Representation, Record, Use; arrows label the verbs encoded, stored, interpreted. The Use card is emphasised in red soft. Bottom band with a red left accent contains four dashed ghost cards aligned under each stage above, naming the metadata facet (Unit, Definition, Collection method, Owner) that protects that stage from meaning loss. Dashed tie lines connect each stage to its facet. OBSERVATION TO USE · FOUR STAGES TIED TO ONE METADATA LAYER 1DMBOK 2 §12ObservationA real-world eventis seen or measured 2ISO 11179-1RepresentationValue, symbol,or category encoded 3DMBOK 2 §12RecordStored with fields,timestamp, source 4UK GDQF Pr.2UseQuestion, action,consequence on the day encoded as stored as interpreted for METADATA TRAVELS WITH EVERY STAGE · ISO/IEC 11179-1:2023 Metadata layer Description of data, registered for reliable reuse UNIT the physical quantity DEFINITION what the value means COLLECTION METHOD how it was captured OWNER who answers for it ransfordsnotes.com

Data only becomes useful when representation, record, and decision use travel together with the metadata that preserves their meaning. ISO/IEC 11179-1:2023 frames metadata as the layer that makes that survival possible.

Context turns a value into a usable data point

The same value 37.2 reads as fever, temperature, or percent depending on what metadata is wrapped around it.

Context turns a value into a usable data point Four stage cards left to right show the same value 37.2 wrapped in successive metadata bands. Stage 1 has the bare red value pill alone. Stage 2 adds an outer dashed Unit band. Stage 3 adds an inner Definition band. Stage 4, emphasised in red soft, adds the outermost Source plus Use band. +1 arrows connect adjacent stages. A red-accent callout below states that without the wrappers, 37.2 could be a clinical fever, a summer afternoon, a battery percentage, or sensor noise. SAME VALUE · DIFFERENT MEANING AT EACH METADATA LAYER 1Bare value37.2The number 37.2.Nothing else known. 2ISO 11179Add UnitUNIT37.237.2 degreesCelsius. 3ISO 11179Add DefinitionUNITDEFINITION37.2A clinical bodytemperature reading. 4UK GDQF Pr.4Add Source + UseUNITDEFINITIONSOURCE / USE37.2Taken at 09:00 byGP for triage. +1 +1 +1 Without the wrappers 37.2 could be a clinical fever, a summer afternoon, a battery percentage, or a sensor noise reading. ransfordsnotes.com

Meaning is added by wrapping a value in metadata, not by changing the value. The same 37.2 reads as fever, temperature, or percent depending on Unit, Definition, Source, and Decision context. ISO/IEC 11179-1:2023 names those wrappers the data element definition, value domain, and conceptual domain.

Where context is dropped: four named failure signatures

Each stage of the observation-to-use journey has a meaning-preserving path and a named failure signature.

Context loss is the first data failure at every stage Four split cards left to right, one per observation-to-use stage. Top zone of each card on white shows the meaning-preserving path SUCCEEDS WITH. A thin red rule splits the card. Bottom zone on red soft names the dropped metadata facet (PURPOSE, UNIT, SOURCE, DEFINITION) and the failure signature FAILS WHEN. Arrows in red connect adjacent cards. A red-accent legend below names the UK Government Data Quality Framework principle that requires teams to document both paths. EACH STAGE HAS A MEANING-PRESERVING PATH AND A FAILURE SIGNATURE 1ObservationSUCCEEDS WITHCaptured withpurpose stated.PURPOSE LOSTFAILS WHENCaptured byaccident, no purpose. 2RepresentationSUCCEEDS WITHValue encodedwith unit and scale.UNIT LOSTFAILS WHENValue encodedwith no unit. 3RecordSUCCEEDS WITHStored with sourceand timestamp.SOURCE LOSTFAILS WHENStored bare,no provenance. 4UseSUCCEEDS WITHRead with the samedefinition as written.DEFINITION DRIFTFAILS WHENRead with adifferent definition. Why both zones matter UK GDQF Principle 5: communicate quality. Consumers need both the happy path and the named failure mode to trust a dataset. ransfordsnotes.com

Every stage of the observation-to-use journey has both a meaning-preserving path and a failure signature. The UK Government Data Quality Framework Principles 1 and 5 require teams to document both so consumers know when the data is fit to use.

Deterministic Data course visual for What data is

Think about it

The National Health Service holds 1.6 billion data items - but cannot answer a simple question without a data standards programme.

In 2022, NHS England published its Data Strategy stating that the health service holds approximately 1.6 billion data items across clinical records, imaging, prescriptions, and operational systems. Yet a GP in Manchester and a consultant in London could look at the same patient record and see different information, because the same field - say, a diagnosis code - was stored using different classification systems across different trusts.

The problem was not volume. The NHS had enormous quantities of data. The problem was that raw data, without shared definitions, agreed formats, and consistent standards, is not information. It is noise at scale. The NHS Data Standards Authority was created in 2020 specifically to address this: to transform 1.6 billion data items from a collection of local facts into a national information asset.

That distinction between data and information is what this module is about.

Defining data

Data is a collection of raw facts represented as symbols, numbers, text, images, or signals. The word comes from the Latin datum, meaning "something given." In computing and information science, data is the lowest layer of recorded observation before any processing or interpretation has taken place.

Data can be a number (37.2), a word ("confirmed"), a pixel value (RGB 255, 0, 128), a sensor voltage reading, or a GPS coordinate pair. What makes something data is not its form but its status: it is a raw, recorded observation. It has no inherent meaning on its own; meaning is assigned through context and interpretation.

The distinction between data, information, and knowledge is often treated as academic, but it has real consequences for data management practice. Organisations routinely store enormous volumes of data while believing they have information. A database of 50 million customer records is data. Only when it has been cleaned, contextualised, and made queryable for a specific purpose does it approach information. Confusing the two leads to overinvestment in storage and underinvestment in processing.

With an understanding of defining data in place, the discussion can now turn to structured, unstructured, and semi-structured data, which builds directly on these foundations.

Structured, unstructured, and semi-structured data

Not all data has the same form. The distinction between these three categories shapes how data is stored, processed, and analysed.

  • Structured data conforms to a predefined schema: rows, columns, and data types are fixed in advance. Relational databases are the canonical storage format. Examples include sales transactions, employee records, and sensor logs with fixed fields.
  • Unstructured data has no predefined schema. It cannot be directly queried with standard database tools without preprocessing. Examples include email bodies, social media posts, audio recordings, CCTV footage, and freeform clinical notes.
  • Semi-structured data has some organisational properties (tags, markers, hierarchy) but does not conform to a rigid relational schema. JSON and XML files are the most common examples. A product catalogue in JSON has consistent key names but variable numbers of attributes per product.

The IDC Data Sphere report of 2023 estimated that the global datasphere would reach approximately 120 zettabytes by the end of 2023. One zettabyte is approximately 1 trillion gigabytes. The majority of this growth is unstructured data: video surveillance, audio streams, social content, and IoT sensor output. Only a small fraction is structured data held in conventional databases.

With an understanding of structured, unstructured, and semi-structured data in place, the discussion can now turn to everyday data sources, which builds directly on these foundations.

“Data is a representation of facts, concepts, or instructions in a formalised manner suitable for communication, interpretation, or processing by human beings or by automatic means.”

ISO/IEC 11179-1:2023, Data element concepts - Section 3.1, Data

Everyday data sources

Data is generated constantly by ordinary systems and activities. Recognising data sources is a practical skill for anyone working with information systems:

  • Form submissions: a user filling in a name and postcode on a government service creates two structured data fields.
  • Sensor readings: a smart electricity meter sends consumption readings every 30 minutes; each reading is a timestamped numeric value.
  • GPS coordinates: location services record latitude, longitude, altitude, and timestamp. A single taxi journey may generate thousands of GPS pings.
  • Web server logs: every HTTP request to a website generates a log entry containing IP address, timestamp, requested URL, and response code.
  • Images and video: a CCTV camera continuously generates pixel arrays. Each frame is a matrix of numeric colour values. Without additional processing (object detection), the raw pixel data has no semantic content.
Check your understanding

A hospital's patient records database contains 50 million rows, each with a patient ID, admission date, diagnosis code, and treatment code. The diagnosis codes use three different coding systems across different hospital sites. Is this data or information?

Common misconception

“A large database of records means an organisation has useful information.”

Volume does not equal value. A database of 50 million customer records is data, not information. Only when it has been cleaned, contextualised, and made queryable for a specific purpose does it approach information. Raw data without cleaning, documentation, and context is expensive to store and difficult to use. The IDC estimate of 120 zettabytes includes a vast proportion that is never accessed after creation. Overinvesting in storage while underinvesting in processing and documentation is one of the most common data strategy errors.

Key takeaways

  • Data is a raw, uninterpreted symbol or signal. It only becomes information when context is applied, and knowledge when that information is interpreted and understood.
  • Structured data conforms to a fixed schema (relational tables); unstructured data has no schema (documents, images, audio); semi-structured data has partial organisation through tags or hierarchy (JSON, XML).
  • Global data volumes are growing rapidly, with the IDC estimating approximately 120 zettabytes as of 2023, predominantly unstructured IoT, video, and social content.
  • Volume does not imply value. Uncleaned, undocumented data imposes storage cost without delivering analytical benefit. The most common data strategy error is overinvesting in storage and underinvesting in processing.

You can now distinguish data from information and classify datasets by structure. The next module examines the DIKW hierarchy in depth: a framework for understanding how raw data transforms into actionable wisdom through context, interpretation, and judgement. The question shifts from "what is data?" to "what do we do with it?"

Standards and sources cited in this module

  1. ISO/IEC 11179-1:2023: Information technology - Metadata registries

    Data element concepts: formal definitions of data, information, and their relationships.

  2. W3C Data on the Web Best Practices (DWBP)

    Section 2: Data concepts including structured/unstructured distinctions and machine-readability.

  3. IDC Data Sphere Report 2023

    Global DataSphere forecast: 120 zettabyte estimate and breakdown by data type.

  4. UK Met Office: How we create a forecast

    Atmospheric data ingestion volumes and the processing chain from raw sensor data to public forecast.

Course overview: Data1 / 26Next: DIKW hierarchy