Definition:
Business Intelligence (BI) is the set of processes, practices, architectures and tools that turns an organisation’s data into consistent information for analysing its activity and supporting decisions. It includes data collection and integration, organisation and modelling, analysis, visualisation and the distribution of reports, indicators and dashboards.
BI can use current and historical information from internal systems or external sources. It is not limited to installing an application: the usefulness of its output depends on data quality, metric definitions, permissions and the context in which the information is interpreted.
Table of contents
How Business Intelligence works
A BI system connects several stages so that operational data can be queried consistently:
- Sources: business applications, databases, spreadsheets, cloud platforms and external sources provide data with different structures and refresh frequencies.
- Integration and preparation: ETL or ELT processes extract, combine, transform and clean data before analysis.
- Organisation and modelling: repositories, semantic models and business rules connect information and define common metrics.
- Analysis and distribution: queries, reports, alerts, visualisations and dashboards enable authorised users to explore and share results.
Early BI environments relied heavily on technical teams to prepare queries and reports. Current platforms include self-service capabilities, but user autonomy requires governed models and shared definitions to prevent incompatible calculations.
BI data and architecture
A Data Warehouse is a common Business Intelligence source because it consolidates information from several systems and retains an analysis-oriented structure. Data marts, data lakes, lakehouse architectures, operational applications or direct connections may also be involved, depending on volume, latency and query requirements.
Hadoop became relevant for processing large distributed datasets and can still form part of particular architectures, but it is not a required BI component. Architecture needs to reflect the sources, refresh frequency, performance, security and operating costs.
Preparation reduces errors and format differences but does not guarantee that information is correct. Data provenance, quality, freshness and access need to be controlled, and the meaning of each indicator needs to be agreed.
Business Intelligence, analytics and Big Data
Business Intelligence and business analytics overlap, but the terms are not fully interchangeable in every context. BI commonly covers the infrastructure, processes and distribution of information used to understand activity. Analytics focuses on methods for describing, diagnosing, predicting or recommending actions.
Data mining is a technique that can be used within a BI initiative to discover patterns. Big Data refers to data and architectures whose scale, speed or variety require particular approaches. An organisation can use BI without a Big Data platform and analyse Big Data outside a BI environment.
Modern platforms may incorporate predictive models, natural language or artificial intelligence. These capabilities extend analysis, but their output needs validation and does not automatically turn a correlation into a causal explanation.
Business Intelligence tools
Tools may cover several layers or specialise in one part of the process:
- Visualisation and self-service: Power BI, Tableau and Qlik Sense enable users to explore data and build interactive reports or dashboards.
- Enterprise reporting and analytics: SAP BusinessObjects, IBM Cognos Analytics and Oracle Analytics provide reporting, distribution and administration capabilities.
- Cloud platforms and semantic layers: Looker, Domo and Strategy One combine source connections, modelling, analysis and collaboration with different scopes.
These examples are not closed categories, and their capabilities overlap. Selection needs to consider supported sources, governance, data visualisation, integration, performance, licensing and the team’s ability to maintain the solution.
Business Intelligence applications and limitations
BI is used to analyse sales, finance, operations, inventory, customer service, campaigns and profitability. In marketing, it can connect investment, response and outcomes, examine groups through market segmentation and support the monitoring of agreed objectives.
A report or dashboard is not in itself a decision or an unquestionable source of truth. Incomplete data, poorly defined metrics, duplication, selection bias or inappropriate permissions can produce misleading conclusions even when the visualisation is technically correct.
Implementation also requires cost, security, privacy, refresh, performance and adoption to be managed. BI creates value when information is integrated into a decision process with accountable owners, criteria and mechanisms for reviewing the results.