Business Intelligence Foundation
An organisation running several systems — an accounting package, an operations system, a spreadsheet somebody maintains — where every management question requires someone to assemble an answer, and two people asked the same question return different numbers.
The problem
The disagreement is not about data. It is that nobody has ever written down what the numbers mean.
In scope
- Source inventory
- Agreed metric definitions
- A consolidated reporting layer
- Dashboards
- Scheduled distribution
- The governance to keep definitions stable
Out of scope
- Replacing any source system.
- Machine learning and prediction — this establishes that the organisation can agree on what happened, which is the prerequisite nobody skips successfully.
- Real-time streaming, which is almost never the actual need.
What this would cover
Grouped by module — open the ones you want to read.
Source inventory
- Every system holding numbers management cares about, what it holds, how it can be read, and how reliable it is
Metric definition
- The core work, and it is organisational rather than technical: what counts as revenue, when is a sale recognised, does a cancelled order count, is a month calendar or operational. Each definition written, agreed, dated and owned by a named person.
Consolidation
- Scheduled extraction into one reporting store, with failed extractions surfaced loudly rather than producing a silently stale dashboard
Dashboards
- By audience: an owner view, a departmental view, a board view; each showing few metrics rather than many
Scheduled distribution
- Reports delivered to where people already read messages, because dashboards that must be visited are not visited
Drill-through
- Every figure traceable to its source rows, which is the only thing that ends arguments about whether a number is right
Definition governance
- A change process, so a metric changing creates a new version rather than silently rewriting last quarter
Data model
- SourceSystem → Extraction → ExtractionLog
- MetricDefinition → MetricVersion → Owner
- Dashboard → Panel
- ScheduledReport
- DataQualityCheck
Invariants
- Every metric has a written definition, a version and a named owner.
- Changing a definition creates a version; history is never silently restated.
- Every displayed figure traces to source rows.
- A failed or stale extraction is shown on the dashboard — a dashboard that looks normal while its data is three days old is actively dangerous.
Offline behavior
Not applicable.
Hard trade-offs
The difficult decisions, stated plainly — not trimmed for length.
The hard part is not technical and it is not Afivox's to solve.
Getting finance and operations to agree on what revenue means, in writing, is a political exercise that can take longer than the build. Afivox facilitates it and documents the outcome, but the organisation has to decide, and a sponsor with authority to settle disputes must be named before the engagement starts. Projects that skip this produce a beautiful dashboard that two departments refuse to accept.
Source data quality caps what is possible, and the cap is usually discovered mid-project.
Phase 1 assesses each source and states plainly what cannot be reported on — sometimes the correct recommendation is to fix a source system first and not to build this yet.
End state
What would be true about their day once this is running.
- Each core metric has one written, owned, versioned definition, and two people asking the same question get the same number.
- Reports arrive where management already reads rather than waiting to be opened.
- Any figure can be drilled to its source.
- When data is stale, the dashboard says so.
This is a deliberate first step toward Enterprise Data & Decision Intelligence Platform — its data model is designed as a genuine subset, so growing into it later doesn't mean starting over.
Let's map how your operation actually runs.
One session. We look at what's breaking, and what we'd build around it — whether or not you hire us afterward.
Start the operations review