Four people. Four answers. All defensible.
Almost never a reporting-tool problem. It is a definition problem wearing a dashboard, and no amount of visualisation fixes it.
Thirty minutes · no predetermined package
What one agreed definition looks like, four ways.
Same records underneath every view. Change the period, switch the question, click a row to isolate it. Nothing here is reconciled by hand.
The first one is not a phase. It is the project.
Most reporting engagements fail before a chart is drawn, and then draw the chart anyway.
Agree the numbers
One written definition per metric, with an owner and a source. Most reporting projects fail here and build anyway.
Not on exports
Reporting that reads the operational database, so the number in the meeting is the number in the system.
When there are many sources
A warehouse and a modelling layer when the data genuinely lives in five systems and reconciliation is the work.
Power BI, Tableau, Looker
Often the right answer. The tool is rarely the constraint; the definitions and the pipeline underneath it are.
What untrustworthy reporting looks like from inside.
Four people, four answers, all defensible.
The monthly pack takes three days to assemble and is out of date when it is presented.
A number is corrected in the meeting, and the correction never reaches the source.
Two dashboards exist for the same question because neither team trusted the other's.
The forecast rests on a pipeline stage that means three different things depending on who moved the record.
Someone exports to a spreadsheet to do the calculation the dashboard was supposed to do.
Six questions before anyone builds a dashboard.
The first one is the one nobody can answer, and it is the reason the rest exist.
Five stages, and the writing comes before the charting.
Write the definitions
one pageEach metric, its formula, its source system and its owner. Signed off by the people who will argue about it later.
Reconcile the sources
the rulesWhere two systems disagree, which wins and why. This is the actual project; the visualisation is the last week.
Model the data
one layerA modelling layer between the raw records and the charts, so a definition change happens in one place rather than nine.
Build the views
by audienceBoard, executive and operational are three different dashboards, not one dashboard with more filters.
Hand over the definitions
maintainableThe document, not just the dashboard. A metric nobody can re-derive is a metric that quietly drifts.
The numbers differ. The disagreement does not.
Lending and financial services
Funded volume, conversion by stage and commission liability, all needing to reconcile with servicing.
Home services
Booked rate by call handler, revenue per technician and maintenance renewal, across dispatch and accounting.
Professional services
Utilisation, realisation and pipeline against delivery capacity, which usually live in three systems.
Manufacturing and distribution
Quote-to-order conversion, margin by product line and inventory turns from an ERP that reports awkwardly.
What people ask about executive dashboards and reporting.
What should an executive dashboard show?
The four or five numbers a decision would actually change, defined once, sourced from the same records the floor uses. Anything beyond that is a maintenance cost with no return.
Why do our dashboards disagree with each other?
Almost always a definition problem rather than a data problem. Two teams counting slightly different things at slightly different times will produce two defensible numbers from the same database.
What is a single source of truth?
One system designated as authoritative for each fact, with every report reading from it rather than from a copy. It is a governance decision before it is a technical one.
Do we need a data warehouse?
Only when the data genuinely lives in several systems and reconciling them is ongoing work. For one or two sources, reporting directly on the operational records is cheaper and fresher.
Should we use Power BI, Tableau or Looker?
Whichever your team already knows. The tool is rarely the constraint — the definitions and the pipeline underneath it are, and those problems follow you to any tool.
What is the difference between an executive and an operational dashboard?
Audience and cadence. Executive answers whether the plan is working, monthly or weekly. Operational answers what to do next, today. Combining them produces something nobody uses.
How often should a dashboard refresh?
As often as a decision could change because of it. Daily is right for most executive reporting; real time is usually a cost with no corresponding decision.
What is data reconciliation?
Deciding which system wins when two disagree, and applying that rule consistently. Without it, every report is an argument waiting to happen.
How long does a reporting project take?
The definitions take longer than the build. Expect the first week to be conversations, and expect those conversations to change what everyone thought they wanted.
Can you build reporting on our existing CRM?
Usually, and it is the cheapest option when there is one main source. Reporting sitting on the operational records is fresher than anything built on a nightly export.
What is board reporting and how is it different?
A narrow set of numbers with commentary, presented monthly or quarterly to people who were not in the operational detail. It needs stability of definition more than it needs depth.
Who should own a dashboard once it is built?
Whoever owns the decision it informs. A dashboard owned by whoever built it stops being maintained the moment they move on.