Numbers everyone can agree on.
Bring the data together, define what each measure means, and make it answer the questions decisions actually depend on.
Useful when two people can pull the same report and get two different answers.
Illustrative system example · fictional product and data · not client work
Decision dashboardPRISM · illustrative system, fictional data
You may need this when
Some of this probably sounds familiar.
- Two departments report the same metric and the numbers do not match.
- The monthly pack takes days of copying and pasting.
- Nobody is quite sure how a headline figure is calculated.
- The information exists somewhere, but not anywhere a decision can use it.
- Reporting stops entirely when one person is on leave.
What we could build
One place the numbers are produced.
Defined once, calculated the same way every time, and traceable back to the source.
- Reporting warehouseOne place the numbers are calculated, rather than five spreadsheets.
- DashboardsThe measures your decisions depend on, not every measure available.
- Data quality checksProblems surfaced before they reach a board pack.
- Automated reportingThe recurring pack that builds and distributes itself.
- Metric definitionsWritten, agreed meanings so a number means one thing.
- Analysis on a real questionA specific answer, with its working shown.
See an example
The same question, three different periods.
Change the period. The measure updates and so does the evidence behind it — because a number without its source is just an assertion.
Every option is shown below. The selector needs JavaScript, so it is not offered here.
OutcomeA monthly figure you can defend, because every line traces to invoices.
OutcomeThe quarterly view shows Enterprise growing where the monthly view suggested North.
OutcomeOver a year the trend reverses: the region that looked strongest is close to flat.
Illustrative interfaces and fictional data. Not client work.
Traceability is the point. Anyone challenging a figure can see which records produced it rather than being asked to trust it.
What you receive
Reporting that runs without you.
- A working data model and the pipelines that keep it current
- Dashboards or reports pointed at the decisions you named
- Written metric definitions agreed with the people who use them
- Data quality checks that flag problems before a person sees them
- Agreed source code, data and infrastructure handed over according to the project scope and terms
- Handover documentation and any agreed support arrangements
Technical depth
Ingestion
Scheduled or event-driven loads from the source systems, with schema checks so a silent format change does not corrupt a report.
Modelling
A warehouse layered from raw through cleaned to reporting models, so a definition changes in one place.
Quality
Row counts, null and range checks, and reconciliation against a trusted source, surfaced as alerts rather than buried.
Lineage
Each reported number traceable to the records and transformations that produced it.
Access
Permissions matched to who should see which figures, rather than one shared login.
Analysis
Statistical methods chosen for the question, with assumptions and limitations stated alongside the result.
Often combined with
What usually sits next to this.
- Forecasting & predictive modelling →When planning needs a view of what may happen next.
- Automation & integrations →When the data has to move reliably before it can be reported.
- AI for business →When people want to ask questions of the data in their own words.
Tell us which number you do not trust.
Naming one disputed figure usually reveals the whole problem.