# Business Intelligence

> Business intelligence from Boltcliq: a trusted semantic layer, natural-language analytics and decision dashboards that connect insight to action across African enterprises and public bodies.

Canonical URL: https://www.boltcliq.com/solutions/business-intelligence
Part of: https://www.boltcliq.com/llms.txt

Analysis at the scale of the whole organization, in the hands of the people who decide.
- Leaders should not wait weeks for a report. Operators get answers in plain language, analysts get a trusted semantic layer and everyone sees one version of the truth.
- One truth: Shared definitions across every team
- Plain language: Ask questions and get verifiable answers
- Decision led: Views built around real decisions
- Insight to action: Analysis connected to workflows
### What we do
- Most organizations have more data than answers. Finance, operations and sales each run a report that is correct by its own definition, and the three do not match. A senior meeting begins with an argument about the numbers. Business intelligence done well ends that argument by agreeing definitions once, putting them in a shared layer and building every view from it.
- We build that layer and the experiences on top. A semantic layer holds metrics such as revenue, active customer, days sales outstanding and portfolio at risk, each with an owner and a calculation. Analysts and tools query it, so a number means the same thing in a spreadsheet, a dashboard and an agent answer. Natural-language analytics lets a manager ask a question and get an answer with the query, the definition and the sources shown.
- We build for decisions rather than for charts. A branch manager needs to know which loans need a visit today. A procurement lead needs to know which suppliers are late. We design views around those decisions and connect them to the next step, whether that is a task, a message or an approved action. Dashboards are the start, not the finish.
### Everyone reading the same numbers
- We often begin with a simple exercise. We ask five teams to calculate the same metric from their own data. The answers differ, sometimes by a lot. We trace each difference to a definition, a timing rule or a data error, and then resolve it with the owner.
- That exercise is the foundation of a trusted layer. After it, a number on a screen can be clicked through to its definition, its calculation and its source. Trust is built by showing the working, not by asserting the answer.
- Metrics with owners and definitions
- Click-through to calculation and source
- Differences resolved once and recorded
### Capabilities
**Semantic layer**
- We define metrics, dimensions and business rules in one governed layer that sits between your data and every analytical tool. Each metric has an owner, a description and a test. The layer works with spreadsheets, BI tools, APIs and agents.
- Governed metrics and dimensions
- Owners, descriptions and tests
- Access from tools, APIs and agents
- Agree the number once and reuse it everywhere.
**Natural-language analytics**
- Managers ask questions in plain language, in English or French and, where models support it, in local languages. The system translates the question into a governed query, runs it and shows the answer with the definitions and sources. Where it is unsure, it asks a clarifying question rather than guessing.
- Questions answered from governed metrics
- Definitions and sources shown
- Clarifying questions when unsure
- Answers people can check will be used.
**Decision dashboards**
- We design views around the decisions people make: who to call, what to approve, where to send stock. Each view shows the few measures that matter, the trend and the next action. We test layouts with the users and remove anything they do not use.
- Built around named decisions
- Next action beside each insight
- Mobile and low-bandwidth variants
- A dashboard no one acts on is decoration.
**Data quality and lineage**
- We add tests for freshness, completeness and plausibility, and show data health on the views themselves. Lineage shows how each number was produced, from source through transformation. Issues route to the owner of the source.
- Freshness and completeness tests
- Lineage from source to metric
- Issue routing to owners
- Data problems should be visible before a meeting.
**Forecasting and advanced analysis**
- We add forecasts and statistical analysis for demand, cash, portfolio risk and capacity. Models use the same governed inputs and publish their error. Analysts can extend them in notebooks and promote the result to the shared layer.
- Forecasts with published error
- Notebooks connected to governed data
- Promotion path to the shared layer
- Looking ahead beats explaining the past.
**From insight to action**
- We connect analysis to workflows. A flagged account creates a task for a relationship manager. An anomaly in stock opens a review. The result of each action returns to the data, so you can learn what works.
- Tasks and approvals from dashboards
- Closed loop from action to outcome
- Audit of who decided what
- Outcomes improve when insight reaches a person who can act.
### Where it applies
- Retail and microfinance
- Branch managers need to know which clients are at risk of missing payments, but reports arrive weeks late and differ by region.
- A governed portfolio-at-risk metric and a daily branch view list accounts to visit, with the reasons and history.
- The branch manager decides whom to visit and what terms to discuss.
- Government revenue
- A revenue authority holds taxpayer, payment and enforcement data in separate systems.
- A semantic layer standardises metrics for collections and compliance, with analysis by sector and region, under strict access control.
- Senior officials decide on policy and enforcement priorities. Officers decide on cases.
- Telecommunications
- Marketing, network and finance teams define churn and active user differently.
- A shared metric layer and plain-language analytics let each team explore the same governed numbers.
- Executives decide on pricing and network investments.
- Healthcare systems
- A health ministry needs facility performance, stock and referral data from many clinics, often reported late.
- A decision dashboard shows stock-outs and delays by district, with data quality indicators.
- District managers decide on resupply and staffing. Clinicians decide on care.
### Built on
**Zeus**
- Provides the semantic layer, natural-language questions and applications built on shared objects.
- Boltcliq product
- Every team uses the same definitions.
**Odin**
- Applies access rules so people and agents see only the rows and fields they should.
- Boltcliq product
- Analytics respect confidentiality.
**Cyclopes**
- Tracks query use, answer quality and cost, so you know what is used and trusted.
- Boltcliq product
- You can retire what nobody uses.
### What we do and do not do
- Align definitions with the people who own them
- Build a governed semantic layer and decision views
- Add natural-language access with sources shown
- Connect analysis to workflows
- Train your analysts to extend the layer
- Build dashboards nobody asked for
- Let a model invent numbers
- Override the judgement of metric owners
- Lock your data inside a proprietary tool
- Expose data beyond the access rules
### Questions we hear
- Do we have to replace our BI tool?
- No. The semantic layer is designed to work with the tools you already use. We replace a tool only when it blocks the outcome.
- How do you stop natural-language answers being wrong?
- The system translates questions into queries over governed metrics, shows the query and definitions, and asks for clarification when unsure. We test on real questions before release and monitor afterwards.
- Which languages are supported?
- English and French are common starting points. Support for local languages depends on current model quality, and we test on your users before promising it.
- How long does it take to align metrics?
- A first set of five to ten metrics can be aligned within weeks. The pace depends on the number of teams and the state of the data.
- How do you handle sensitive data?
- Access is controlled by role and purpose at row and field level. Which laws apply depends on your sector and country, so confirm with counsel.