Analytics and Visualisation

Analytics and visualisation platforms transform raw data into interactive dashboards, reports, and exploratory views that support operational decision-making. These tools connect to databases, data warehouses, and APIs to present aggregated metrics, trends, and drill-down analysis to staff across technical and non-technical roles.

This page covers general-purpose business intelligence and visualisation platforms. It excludes specialised monitoring tools (see Security and Monitoring), geospatial analysis platforms (see GIS and Mapping), and programme-specific M&E platforms (see M&E Platforms).

Requirements taxonomy

This taxonomy defines evaluation criteria for analytics and visualisation tools. Requirements are organised by functional area and weighted by typical priority for mission-driven organisations.

Functional requirements

Core capabilities that define what the tool must do.

Data connectivity

Data preparation

Visualisation capabilities

Dashboard composition

Sharing and collaboration

Administration

Technical requirements

Infrastructure, architecture, and deployment considerations.

Deployment and hosting

Scalability and performance

Integration architecture

Security requirements

Security controls and data protection capabilities.

Authentication and access control

Data protection

Compliance and certifications

Operational requirements

Administration, monitoring, and maintenance considerations.

Administration

Monitoring and observability

Backup and recovery

Support and maintenance

Commercial and contractual requirements

Pricing, licensing, and vendor considerations.

Pricing and licensing

Vendor assessment

Accessibility requirements

Comparison matrices

The following matrices compare assessed tools using standardised ratings:

Functional capability matrix

Data connectivity

Assessment notes:

• Superset supports 40+ databases via SQLAlchemy; file support requires database loading first
• Grafana excels at time-series databases (Prometheus, InfluxDB) but has limited relational database features
• Metabase provides native connectors for major databases with strong schema exploration
• JupyterLab requires Python libraries (pandas, SQLAlchemy) for database connectivity
• Power BI and Tableau have the broadest native connector ecosystems

Data preparation

Assessment notes:

• Superset’s SQL Lab provides excellent query authoring; semantic layer is available but less mature than commercial options
• Grafana focuses on query transformation rather than visual data preparation
• Metabase’s visual query builder is strong for non-technical users
• JupyterLab uses Python/pandas for transformation, which provides unlimited flexibility but requires coding
• Power BI’s semantic model (formerly Power Pivot) and Tableau’s data model provide mature semantic layers

Visualisation capabilities

Assessment notes:

• Superset includes 50+ chart types including deck.gl geographic visualisations
• Grafana specialises in time-series with strong panel ecosystem
• Metabase has fewer chart types but covers common use cases well
• JupyterLab with Plotly, Altair, or matplotlib provides extensive visualisation options
• Power BI and Tableau have the most polished visualisation experiences

Sharing and collaboration

Assessment notes:

• Superset’s alerting requires Celery worker configuration
• Grafana alerting is mature for infrastructure monitoring use cases
• Metabase embedding and alerting require Pro or Enterprise licence
• JupyterLab focuses on notebooks rather than dashboard sharing; JupyterHub provides sharing infrastructure
• Power BI and Tableau have the most complete collaboration features

Technical capability matrix

Deployment options

Assessment notes:

• Superset cloud hosting primarily via Preset.io (commercial)
• Grafana Cloud is a mature managed service with global regions
• Metabase Cloud offers managed hosting in US and EU regions
• JupyterHub provides self-hosted multi-user deployment; various cloud providers offer managed JupyterLab
• Power BI is primarily cloud-native; Power BI Report Server provides limited on-premises option
• Tableau Server provides on-premises; Tableau Cloud is managed service

Self-hosted infrastructure requirements

API capabilities

Security capability matrix

Authentication methods

Assessment notes:

• Superset uses Flask-AppBuilder for authentication; MFA requires external IdP
• Grafana SAML and LDAP require Enterprise licence; OAuth is available in OSS
• Metabase SSO features require Pro or Enterprise
• JupyterHub provides multi-user authentication; JupyterLab alone has limited auth
• Power BI uses Azure AD exclusively
• Tableau provides comprehensive authentication options across editions

Data protection features

Assessment notes:

• Superset RLS uses SQL-based policies configurable per dataset
• Grafana focuses on dashboard-level access rather than data-level security
• Metabase sandboxing (data access restrictions) requires Pro or Enterprise
• JupyterLab security depends on notebook server and kernel configuration
• Power BI and Tableau provide the most comprehensive data security features

Compliance certifications (cloud/SaaS offerings)

Assessment notes:

• Self-hosted deployments inherit organisation’s compliance posture
• Preset.io (commercial Superset) provides enterprise compliance features
• Grafana Enterprise Cloud offers HIPAA BAA
• Power BI and Tableau have the most comprehensive compliance coverage

Commercial comparison matrix

Pricing models

Cloud pricing estimates (January 2026)

Notes:

• Prices are indicative; verify current pricing with vendors
• Microsoft and Salesforce (Tableau) offer nonprofit programmes with significant discounts
• Self-hosted FOSS options have infrastructure costs but no licence fees

Vendor details

Jurisdictional considerations

Assessment notes:

• All US-headquartered cloud providers are subject to CLOUD Act data requests regardless of data location
• Self-hosted deployment of FOSS tools eliminates jurisdictional concerns
• Power BI and Tableau offer data residency options but remain US-jurisdiction companies

Accessibility matrix

Assessment notes:

• Commercial tools (Power BI, Tableau) have invested more heavily in accessibility compliance
• FOSS tools have active accessibility improvements but lack formal VPAT documentation
• Grafana publishes a VPAT for Grafana Enterprise

Tool assessments

Apache Superset

Type: Open source business intelligence platform Licence: Apache 2.0 Current version: 6.0.0 (December 2025) Repository: https://github.com/apache/superset Documentation: https://superset.apache.org/docs/

Apache Superset is a modern data exploration and visualisation platform originally developed at Airbnb and donated to the Apache Software Foundation in 2021. The platform provides a web-based interface for creating charts and dashboards from SQL databases without requiring users to write code, while also offering a full SQL IDE for advanced users.

Superset connects to any database with a SQLAlchemy-compatible driver, providing native support for 40+ database technologies including PostgreSQL, MySQL, ClickHouse, BigQuery, Snowflake, and Trino. The semantic layer allows defining reusable metrics and dimensions that business users access through a visual query builder.

Key strengths:

The SQL Lab feature provides a comprehensive query authoring environment with syntax highlighting, autocomplete, query history, and the ability to save queries as virtual datasets for visualisation. Query results cache automatically, reducing database load for repeated queries.

The visualisation library includes 50+ chart types, including deck.gl-powered geographic visualisations, pivot tables, and custom plugin support. The chart builder interface uses drag-and-drop metric and dimension selection that generates optimised SQL queries.

Role-based access control operates at both the database and dataset level, with row-level security policies defined in SQL. This allows restricting data visibility based on user attributes without duplicating datasets.

The dashboard filter system supports cross-filtering between charts, filter hierarchies, and filter dependencies. Dashboards can be embedded in external applications using configurable authentication.

Key limitations:

The learning curve for administrators is substantial. Configuration requires understanding of Python, Flask, and SQLAlchemy. Default deployments need additional components (Redis, Celery) for caching and alerting functionality.

Real-time data capabilities are limited compared to Grafana. While query results refresh on schedule, true streaming dashboard updates require external tooling.

Documentation quality varies across features. Core functionality is well-documented, but advanced configurations and troubleshooting often require community forum research.

Mobile responsiveness is basic. Dashboards scale to smaller screens but lack device-specific layout optimisation.

Deployment and operations:

Self-hosted deployment requires Python 3.9+, a metadata database (PostgreSQL recommended for production), and optionally Redis for caching and Celery for async tasks. Official Docker images and Helm charts simplify container deployments. A minimal production deployment runs on 4 GB RAM and 2 vCPUs; typical production environments use 8 GB RAM.

The application initialises its metadata database on first run. Schema migrations occur automatically during upgrades, though complex deployments benefit from manual migration testing.

Preset.io offers managed Superset hosting with enterprise features including SCIM provisioning, enhanced security controls, and dedicated support.

Integration capabilities:

The REST API follows OpenAPI specification with Swagger documentation available at /swagger/v1. API coverage includes dashboards, charts, datasets, databases, and queries. Authentication uses session cookies or OAuth tokens.

Embedding uses iframe with configurable authentication. Guest tokens provide time-limited access to specific dashboards without requiring user accounts.

Export/import functionality supports JSON format for dashboards and datasets, enabling content migration between environments.

Security assessment:

Authentication delegates to Flask-AppBuilder, supporting local accounts, LDAP, OAuth providers (Google, GitHub, Azure AD, Okta), and SAML. MFA requires external identity provider configuration.

Authorisation uses a role-based model with predefined roles (Admin, Alpha, Gamma, Public) and custom role creation. Row-level security policies apply SQL WHERE clauses based on user attributes.

Audit logging captures query execution, dashboard access, and administrative actions. Logs output to standard Python logging infrastructure.

Cost analysis:

Self-hosted Superset has no licence cost. Infrastructure costs for a typical organisation:

Preset managed hosting starts at approximately $50/user/month with volume discounts.

Organisational fit:

Best suited for: Organisations with Python/SQL expertise seeking full control over their BI stack; teams with existing data warehouses wanting self-service analytics; organisations requiring on-premises deployment for data sovereignty.

Less suitable for: Teams without technical capacity to manage infrastructure; organisations needing polished mobile experiences; use cases requiring real-time streaming dashboards.

Grafana

Type: Open source analytics and monitoring platform Licence: AGPL 3.0 (core), proprietary (Enterprise features) Current version: 12.2.1 (January 2026) Repository: https://github.com/grafana/grafana Documentation: https://grafana.com/docs/

Grafana is a multi-platform open source analytics and interactive visualisation application. While originally designed for infrastructure monitoring with time-series databases, Grafana has expanded to support general analytics use cases including business metrics, application monitoring, and observability across diverse data sources.

The platform excels at real-time dashboards with automatic refresh, alerting on metric thresholds, and correlation of data across multiple sources. Grafana connects natively to time-series databases (Prometheus, InfluxDB, Graphite), relational databases (PostgreSQL, MySQL), cloud monitoring services (CloudWatch, Azure Monitor), and through plugins to 150+ additional data sources.

Key strengths:

Dashboard composition is highly flexible with a responsive grid layout, template variables for dynamic filtering, and annotations that overlay events on time-series visualisations. Panels support drill-down links to other dashboards or external systems.

The alerting system evaluates conditions across any data source and delivers notifications through email, Slack, PagerDuty, webhooks, and 30+ notification channels. Alert rules support multi-dimensional conditions with templated messages.

Plugin architecture extends every aspect of the platform. The plugin ecosystem includes data source plugins, panel plugins, and app plugins that bundle related functionality. Organisations can develop private plugins using the documented SDK.

Grafana Loki (logs), Tempo (traces), and Mimir (metrics) integrate natively, providing a unified observability stack. This combination suits organisations standardising on the Grafana ecosystem.

Key limitations:

Traditional BI capabilities are weaker than purpose-built BI tools. Pivot tables, complex calculated fields, and semantic modelling require workarounds or are unsupported.

SQL database support, while functional, lacks the query building and data modelling features of BI-focused tools. Users comfortable with SQL can work effectively, but non-technical users may struggle.

The AGPL licence requires careful consideration. Modifications to Grafana source code must be published, which affects organisations wanting to customise the platform without open-sourcing changes.

Row-level security is minimal in Grafana OSS. Folder-level and dashboard-level permissions are available, but restricting data visibility based on user attributes requires Enterprise features or data-source-level controls.

Deployment and operations:

Grafana runs as a single binary with minimal dependencies. A minimal installation needs 512 MB RAM; production deployments benefit from 2+ GB RAM depending on concurrent user load.

The embedded SQLite database suffices for small deployments. Production environments should use PostgreSQL or MySQL for the configuration database to enable high availability.

Official Docker images, Helm charts, and Linux packages simplify deployment. Grafana supports horizontal scaling behind a load balancer when using an external database.

Grafana Cloud provides fully managed hosting with generous free tier (10 users, 10K metrics series, 50 GB logs). Cloud tiers include additional features like machine learning-powered alerting and enhanced security.

Integration capabilities:

The HTTP API provides comprehensive coverage for dashboards, alerts, users, and administration. Authentication uses API keys or OAuth tokens. Service accounts (non-human identities) support automation scenarios.

Terraform provider enables infrastructure-as-code management of dashboards, alerts, and data sources. The provider covers most Grafana resources and integrates with CI/CD pipelines.

Embedding uses iframe with authentication via direct access tokens, anonymous access, or authentication proxy. Grafana Enterprise adds signed URLs for secure embedding.

Security assessment:

Authentication supports local accounts, OAuth (generic and provider-specific for Azure AD, GitHub, Google, Okta), LDAP, and SAML. MFA is available for local accounts. SAML and LDAP require Enterprise licence.

Authorisation uses organisation-scoped roles (Admin, Editor, Viewer) with folder and dashboard permissions. Team-based permissions group users for access management. Enterprise adds data source permissions and enhanced RBAC.

Audit logging in Enterprise records user actions, API calls, and administrative changes. OSS version provides basic access logs via standard web server logging.

Cost analysis:

Self-hosted Grafana OSS has no licence cost:

Grafana Cloud pricing:

• Free: 10 users, limited metrics and logs
• Pro: $50/user/month (minimum 5)
• Advanced: $100/user/month
• Enterprise: Custom pricing

Nonprofit discounts available through Grafana for Good programme.

Organisational fit:

Best suited for: Technical teams monitoring infrastructure and applications; organisations with time-series data (IoT, performance metrics); teams standardising on the Grafana observability stack.

Less suitable for: Traditional business reporting needs; non-technical users needing self-service analytics; organisations requiring semantic layer and data modelling capabilities.

Metabase

Type: Open source business intelligence platform Licence: AGPL 3.0 (Open Source), proprietary (Pro and Enterprise) Current version: 0.58.2 (January 2026) Repository: https://github.com/metabase/metabase Documentation: https://www.metabase.com/docs/

Metabase is designed for self-service business analytics, enabling non-technical users to ask questions of data without writing SQL. The platform emphasises ease of use with a visual query builder that generates SQL, automatic data discovery that surfaces interesting insights, and minimal configuration requirements for deployment.

The visual query builder allows selecting tables, picking columns, applying filters, and choosing aggregations through point-and-click interaction. The generated SQL is visible and editable, providing a learning path for users interested in developing SQL skills.

Key strengths:

The query builder genuinely enables non-technical users to explore data independently. Automatic join detection, intelligent column suggestions, and natural language question parsing (experimental) reduce the barrier to data exploration.

Setup simplicity is notable. A single JAR file or Docker container connects to databases and serves dashboards within minutes. The initial configuration wizard guides administrators through database connection and user setup.

The native question saving creates a curated collection of common queries that serve as starting points for exploration. Questions can be organised in collections with access controls, creating a knowledge base of analytical queries.

X-ray analysis automatically generates dashboards summarising tables, fields, and segments. This accelerates initial exploration of unfamiliar databases and surfaces patterns that might not be immediately apparent.

Key limitations:

Advanced visualisation capabilities lag behind competitors. Chart types cover common needs but lack the variety and customisation depth of Superset, Tableau, or Power BI.

Performance with large datasets (10M+ rows) requires careful query design. Without extract creation, Metabase queries the source database directly, which can stress production databases. Pro edition adds query caching and result persistence.

The semantic layer is less developed than enterprise BI tools. While models allow defining transformations, the lack of a comprehensive metrics layer means calculated fields often need redefinition across questions.

Single-node architecture in OSS limits scalability. High availability and horizontal scaling require Enterprise edition.

Deployment and operations:

Metabase ships as a Java application running on JVM 11+. The embedded H2 database enables quick start but is unsuitable for production; PostgreSQL or MySQL should be configured for durable deployments.

Resource requirements are modest: 2 GB RAM minimum, 4 GB recommended. The application handles moderate concurrent usage (10-30 simultaneous users) on a single 2-vCPU instance.

Docker images and JAR downloads are available from official channels. Kubernetes deployment requires external database and session storage configuration.

Metabase Cloud offers managed hosting with automatic updates, backups, and monitoring. Regions include US and EU (Ireland).

Integration capabilities:

The API covers questions, dashboards, users, and collections. Authentication uses session tokens obtained via login endpoint or API keys (Pro/Enterprise). Documentation is functional but less comprehensive than competitors.

Embedding supports iframe with signed tokens (Pro/Enterprise) or public sharing links. The embedding SDK provides React components for deeper integration.

Export supports CSV and Excel formats. Native integrations include Slack notifications and Google Drive export.

Security assessment:

Authentication supports local accounts, LDAP (Enterprise), Google OAuth, and SAML (Enterprise). SSO requires paid editions.

Data sandboxing (Enterprise) restricts data access based on user attributes, functioning similarly to row-level security. Sandboxes apply to questions, preventing users from seeing data outside their permitted scope.

Audit logging (Enterprise) records user activity, queries executed, and administrative changes.

Cost analysis:

Self-hosted OSS has no licence cost:

Commercial editions:

• Metabase Cloud Starter: $85/month (5 users, 10 questions)
• Metabase Cloud Pro: Custom pricing
• Self-hosted Pro: $500/month (10 users)
• Self-hosted Enterprise: Custom pricing

Nonprofit discount: 50% off self-hosted Pro and Enterprise.

Organisational fit:

Best suited for: Organisations wanting to democratise data access for non-technical staff; teams needing quick deployment with minimal IT overhead; small to medium scale analytics (under 100 concurrent users).

Less suitable for: Large-scale enterprise deployments; advanced visualisation requirements; organisations needing real-time or streaming analytics.

JupyterLab

Type: Open source interactive development environment Licence: BSD-3-Clause Current version: 4.5.0 (December 2025) Repository: https://github.com/jupyterlab/jupyterlab Documentation: https://jupyterlab.readthedocs.io/

JupyterLab is the next-generation web-based interface for Project Jupyter, providing a flexible environment for interactive computing with notebooks, code, and data. Unlike purpose-built BI tools, JupyterLab is a general-purpose platform that enables analytics through Python, R, or Julia code execution.

The notebook interface combines code cells, output cells, and markdown documentation in a single document. This literate programming approach allows analysts to explain their methodology alongside the analysis, creating reproducible analytical reports.

Key strengths:

Analytical flexibility exceeds any BI tool. Python’s data ecosystem (pandas, NumPy, scikit-learn, PyTorch) enables statistical analysis, machine learning, and custom computations that fixed BI interfaces cannot match.

Visualisation libraries (matplotlib, Plotly, Altair, Bokeh, seaborn) provide publication-quality graphics with complete customisation control. Interactive widgets enable dashboard-like functionality within notebooks.

Reproducibility is inherent to the notebook format. Analysis steps are explicit, versioned with Git, and re-executable. This matters for audit trails, methodology documentation, and collaborative refinement.

Extension architecture supports custom integrations. Extensions add version control interfaces, database explorers, diagram editors, and integrations with cloud services.

Key limitations:

Non-technical user accessibility is fundamentally limited. Using JupyterLab requires writing code. While basic data exploration can use simple pandas operations, the platform is unsuitable for business users without programming experience.

Dashboard creation requires additional tools. Voilà converts notebooks to dashboards, Panel enables interactive applications, but these add complexity and lack the polish of purpose-built dashboard tools.

Multi-user deployment requires JupyterHub, which adds significant operational complexity. Authentication, resource management, and user isolation need explicit configuration.

Security model differs from traditional applications. Notebook servers execute arbitrary code, which creates different threat vectors than rendering dashboards. Proper deployment requires careful isolation.

Deployment and operations:

Single-user JupyterLab installs via pip or conda with minimal resources (1 GB RAM sufficient for basic use). More complex analyses with large datasets require proportionally more memory.

JupyterHub provides multi-user deployment with pluggable authentication (OAuth, LDAP), spawner backends (local processes, Docker, Kubernetes), and resource quotas. Kubernetes deployment using Zero to JupyterHub guide provides scalable infrastructure.

Cloud-hosted options include Google Colab, Microsoft Azure Notebooks, Amazon SageMaker Studio, and various managed JupyterHub providers. These simplify operations at the cost of control and potentially data residency.

Integration capabilities:

JupyterLab connects to any data source with Python or R libraries. SQLAlchemy provides database connectivity; pandas reads files; requests handles APIs. This programmable connectivity exceeds any fixed connector set.

APIs via jupyter-server enable programmatic control. nbconvert exports notebooks to HTML, PDF, and other formats. Papermill enables parameterised notebook execution for automation.

Git integration via jupyterlab-git extension provides version control UI. Real-time collaboration via jupyter-collaboration enables Google Docs-style concurrent editing.

Security assessment:

JupyterLab authenticates via JupyterHub, which supports OAuth (GitHub, GitLab, Google, Azure AD, generic), LDAP, PAM, and custom authenticators. Single-user deployment uses token authentication.

Authorisation in JupyterHub controls server access and admin privileges. File-level permissions follow server filesystem. Code execution permissions are inherent to having server access.

Audit logging depends on JupyterHub and server configuration. Access logs capture authentication events; execution logs require custom configuration or integration with workflow tools.

Cost analysis:

Self-hosted JupyterLab/Hub has no licence cost:

Cloud options vary widely based on compute usage. Google Colab Pro costs $10/month per user; enterprise offerings range higher.

Organisational fit:

Best suited for: Data science teams performing statistical analysis and machine learning; research organisations needing reproducible methodology documentation; technical analysts comfortable with Python or R.

Less suitable for: Business users without programming skills; organisations needing self-service analytics for non-technical staff; scenarios requiring polished, shareable dashboards.

Microsoft Power BI

Type: Commercial business intelligence platform Licence: Proprietary Current version: January 2026 release (monthly updates) Documentation: https://learn.microsoft.com/en-us/power-bi/

Microsoft Power BI is an enterprise business intelligence platform providing self-service analytics, governed data sharing, and embedded analytics. The platform integrates tightly with the Microsoft ecosystem (Azure, Microsoft 365, Dynamics 365) while supporting connections to hundreds of external data sources.

Power BI consists of Power BI Desktop (Windows authoring tool), Power BI Service (cloud-hosted collaboration), Power BI Mobile (iOS and Android apps), and Power BI Report Server (on-premises option). The platform recently integrated with Microsoft Fabric, expanding into data engineering, data science, and real-time analytics.

Key strengths:

The semantic modelling layer (formerly Power Pivot) provides a mature framework for defining relationships, hierarchies, calculated measures, and business logic. DAX (Data Analysis Expressions) enables sophisticated calculations while remaining accessible through formula autocomplete and documentation.

Microsoft 365 integration enables natural workflows. Excel users can connect to Power BI datasets directly. Teams embedding allows dashboard viewing in channel conversations. SharePoint integration provides content management context.

Natural language queries via Q&A allow users to type questions like “sales by region last quarter” and receive automated visualisations. While imperfect, this accelerates exploration for users unfamiliar with the data model.

Dataflows enable cloud-based data preparation with scheduled refresh, reducing the need for separate ETL infrastructure. Power Query provides visual transformation capabilities accessible to non-technical users.

Key limitations:

Windows dependency for authoring limits cross-platform teams. Power BI Desktop runs only on Windows; web authoring covers basic scenarios but lacks full capability parity.

Pricing complexity creates budget uncertainty. Workspace capacity requirements, premium per-user licensing, and add-on features make total cost difficult to estimate without detailed sizing.

The Microsoft ecosystem assumption permeates the product. Organisations not using Azure AD face friction in authentication. Non-Microsoft data sources work but lack the optimisation of first-party connections.

On-premises option (Power BI Report Server) has significant feature gaps compared to the cloud service and receives less frequent updates.

Deployment and operations:

Power BI is primarily a cloud service. Administrators configure tenants through the Power BI admin portal and Azure AD. No infrastructure management is required for the cloud service.

Power BI Report Server provides on-premises deployment for organisations with cloud restrictions. Requirements include Windows Server 2016+ and SQL Server 2016+. Minimum hardware is 8 GB RAM and 4 cores; production environments need significantly more depending on usage.

Microsoft Fabric unifies Power BI with data engineering capabilities in a single SaaS platform with capacity-based pricing.

Integration capabilities:

REST APIs cover administration, embedding, datasets, and push data. Authentication uses Azure AD OAuth 2.0. API capabilities have expanded significantly with Fabric integration.

Embedding options include secure embedding in organisational apps, publish-to-web for public content, and the Power BI Embedded service for ISV scenarios. React and JavaScript SDKs support custom integration.

Power Automate integration enables workflow triggers based on data alerts. SharePoint, Teams, and other Microsoft services have native embedding support.

Security assessment:

Authentication uses Azure AD exclusively. This provides enterprise features (conditional access, MFA, group-based licensing) but requires Azure AD presence.

Row-level security (RLS) uses DAX expressions to filter data based on user identity. Combined with object-level security, this enables comprehensive data access control.

Microsoft’s compliance portfolio applies to Power BI: SOC 2 Type II, ISO 27001, GDPR, HIPAA, and FedRAMP certifications cover the platform.

Cost analysis:

Power BI licensing (January 2026):

Microsoft 365 E5 includes Power BI Pro. The Nonprofits programme provides 10 donated Pro licences and discounted additional licences.

Organisational fit:

Best suited for: Microsoft-centric organisations; enterprises with Azure AD infrastructure; teams wanting integrated experience across Microsoft 365; organisations requiring advanced security and compliance certifications.

Less suitable for: Organisations without Azure AD; Mac-only environments; teams preferring open source tooling; organisations concerned about US jurisdiction for cloud data.

Tableau

Type: Commercial business intelligence platform Licence: Proprietary Current version: 2025.1 (rolling updates) Documentation: https://help.tableau.com/

Tableau is an enterprise analytics platform acquired by Salesforce in 2019. The platform emphasises visual analytics with a drag-and-drop interface that enables rapid exploration and visualisation creation. Tableau pioneered the modern self-service BI market and maintains strong capabilities in data visualisation and exploratory analysis.

The product line includes Tableau Desktop (authoring tool for Windows and Mac), Tableau Cloud (SaaS platform), Tableau Server (self-hosted option), and Tableau Public (free public sharing platform).

Key strengths:

Visual exploration capabilities set the industry standard. The drag-and-drop interface with immediate visual feedback enables rapid hypothesis testing. Automatic visualisation recommendations (Show Me) suggest appropriate chart types based on data characteristics.

Visualisation quality exceeds most competitors. Default formatting produces publication-ready charts. Fine-grained control over every visual element enables pixel-precise customisation.

Tableau Prep provides visual data preparation with joining, pivoting, and cleaning operations. The interface shows data flow and transformations, making complex preparation accessible to non-technical users.

Cross-platform desktop application runs on Windows and Mac with consistent functionality. This addresses Power BI’s Windows-only limitation.

Key limitations:

Salesforce integration emphasis has increased since acquisition. Native CRM Analytics connectivity, Einstein AI features, and branding alignment assume Salesforce ecosystem presence.

Pricing is among the highest in the BI market. Creator licences for full functionality cost significantly more than competing products.

The data engine (Hyper) requires data extraction for optimal performance. Live connections to some databases incur query limitations. Extract refresh management adds operational overhead.

Learning curve for advanced features is substantial. While basic visualisation is intuitive, calculated fields, level of detail expressions, and advanced analytics require significant learning investment.

Deployment and operations:

Tableau Cloud is a fully managed SaaS platform. Administrators configure sites, projects, and permissions through web interface. No infrastructure management required.

Tableau Server deploys on Windows or Linux. Requirements: 8 cores and 32 GB RAM minimum; production deployments typically need 64+ GB RAM. The installer handles most configuration; clustering for high availability requires careful planning.

Tableau Bridge connects Tableau Cloud to on-premises data sources, enabling cloud deployment while keeping data on-premises.

Integration capabilities:

REST API covers sites, projects, workbooks, views, and users. Metadata API provides GraphQL interface for data lineage. Authentication uses personal access tokens or session-based authentication.

Embedding SDK supports JavaScript integration with event handling and filtering. Connected Apps provide secure embedding with JWT authentication. Embedding is available across all paid tiers.

Tableau Extensions API enables custom dashboard components. Extensions can connect to external services, adding functionality beyond built-in capabilities.

Security assessment:

Authentication supports local accounts, Active Directory, SAML 2.0, and OpenID Connect. Site-level authentication configuration allows different methods per site.

Row-level security through user filters applies data restrictions based on logged-in user. Entitlement tables provide scalable permission management for large user populations.

Tableau Trust portal documents SOC 2 Type II, ISO 27001, GDPR compliance, and HIPAA eligibility with BAA availability.

Cost analysis:

Tableau licensing (January 2026):

Tableau Enterprise bundles additional features (Data Management, Advanced Management) at higher per-user rates.

Salesforce.org offers Tableau through the Nonprofit Cloud with donated and discounted licences.

Organisational fit:

Best suited for: Organisations prioritising visualisation quality and exploratory analysis; Salesforce ecosystem users; teams with dedicated analysts who will master advanced features; enterprises needing mature governance capabilities.

Less suitable for: Budget-constrained organisations; teams wanting open source solutions; simple reporting needs that don’t justify the cost; organisations preferring self-hosted FOSS for data sovereignty.

Selection guidance

Decision framework

The following decision tree provides initial guidance based on primary requirements:

                           +------------------+
                           | Primary need?    |
                           +--------+---------+
                                    |
         +--------------------------+---------------------------+
         |                          |                           |
         v                          v                           v
+--------+--------+        +--------+--------+         +--------+--------+
| Infrastructure  |        | Business        |         | Advanced        |
| monitoring      |        | reporting       |         | analytics       |
+-----------------+        +-----------------+         +-----------------+
         |                          |                           |
         v                          v                           v
     Grafana               +--------+--------+          JupyterLab
                           | Technical       |          (with coding)
                           | capacity?       |
                           +--------+--------+
                                    |
              +---------------------+---------------------+
              |                                           |
              v                                           v
    +---------+----------+                    +-----------+---------+
    | Self-hosted OK?    |                    | Minimal IT          |
    +--------------------+                    | Metabase Cloud      |
              |                               +---------------------+
    +---------+----------+
    | Yes     |    No    |
    +---------+----------+
         |          |
         v          v
    Superset    Power BI (Microsoft shop)
    Metabase    Tableau (Salesforce shop)

Recommendations by context

For organisations with minimal IT capacity

Primary recommendation: Metabase Cloud or Power BI (with Microsoft 365)

Organisations with limited technical staff benefit from managed services that minimise operational burden. Metabase Cloud provides a straightforward self-service analytics experience with minimal configuration. If Microsoft 365 is already in use, Power BI Pro (included in E5 or available as add-on) integrates naturally with existing infrastructure.

Alternative: Grafana Cloud for organisations with infrastructure monitoring needs

Avoid: Self-hosted Superset (requires substantial configuration expertise)

For organisations with established IT capacity

Primary recommendation: Apache Superset (self-hosted) or Metabase (self-hosted with Pro features)

Organisations with dedicated IT staff can leverage self-hosted open source solutions for maximum control and cost efficiency. Superset provides deeper SQL capabilities and more visualisation options; Metabase emphasises accessibility for non-technical users.

Alternative: Tableau Server or Power BI Premium for organisations with enterprise requirements and budget

Considerations: Self-hosted deployment requires ongoing maintenance, security patching, and capacity planning. Factor staff time into total cost of ownership calculations.

For organisations with data sovereignty requirements

Primary recommendation: Self-hosted FOSS (Superset, Grafana, or Metabase OSS)

Organisations operating in high-risk environments, handling sensitive beneficiary data, or subject to data localisation requirements should prioritise self-hosted deployment. FOSS options eliminate vendor data access and jurisdictional concerns.

Alternative: Metabase Cloud (EU region) or Grafana Cloud (EU region) provide managed options with European data residency, though US-headquartered companies remain subject to CLOUD Act.

Avoid: Cloud services from US-headquartered vendors when handling highly sensitive data (protection cases, whistleblower information) in jurisdictions with adversarial relationships.

For technical data science teams

Primary recommendation: JupyterLab with JupyterHub

Teams with Python proficiency gain maximum flexibility from notebook-based analysis. Combine with Voilà or Panel for dashboard sharing needs. Integration with machine learning workflows and version control provides reproducibility advantages.

Alternative: Superset for SQL-heavy workflows with visualisation needs

Considerations: Notebook-based approaches require programming skills from all users. Consider Metabase or similar for broader organisational access.

Migration paths

Resources and references

Official documentation

Open source tools

Commercial tools

Relevant standards

See also

Internal documentation relevant to analytics and visualisation tool selection:

• Analytics Strategy -Developing an organisational analytics approach
• BI Platform Implementation -Implementing the selected platform
• Dashboard Design -Creating effective dashboards
• Data Visualisation Standards -Visual design guidelines
• Data Governance Framework -Governance context for analytics
• Security and Monitoring -Related monitoring tools comparison
• M&E Platforms -Programme-specific analytics tools