Build it with AI
Build a business dashboard around the decisions leadership needs to make.
Create a tailored view of revenue, pipeline, operations, customers, workload, or other business KPIs.
Introduction
What a business dashboard has to hold.
Most teams end up with a business dashboard the same way: a board pack assembled from several systems in the week before each meeting. The pack works while the business is simple enough that the assembler understands every number in it. It fails when the assembler becomes a router between five system owners, and every question in the meeting becomes an action item for next month.
Assembling the pack is a week of work, so it answers the questions asked last quarter rather than this one. There is no definition of each headline metric that the whole leadership team has agreed, no lineage from a board number back to the records beneath it, and no way to answer a follow-up question inside the meeting.
What follows covers building a business dashboard: the records it holds (revenue, pipeline, delivery, cash, headcount, and the operating metrics behind each), the systems it reads (Google Sheets and HubSpot), and what it does not fix.
The problem
A board pack assembled by hand from five systems.
Every operational system reports on its own domain with its own definitions, and the pack reconciles them in a spreadsheet whose logic exists only in the assembler’s head. BI tools chart whatever is loaded and are silent on whether the definitions agree.
The records are revenue, pipeline, delivery, cash, headcount, and the operating metrics behind each, and the authoritative copy of most of them already lives in Google Sheets or HubSpot. Revenue in the board pack, revenue in the billing system, and revenue in the CRM are three different numbers, and which is quoted depends on who assembled the slide.
The cost is not the inconvenience: decisions wait for the next reporting cycle.
You're likely here because
- Every question in the leadership meeting becomes an action item
- Assembling the pack is a week of work, so it answers the questions asked last quarter rather than this one.
- When it is wrong, decisions wait for the next reporting cycle
What gets built
Launch builds it, Grow operates it.
Built in Launch
- • Executive KPIs
- • Trend views
- • Drill-down tables
Operated through Grow
- • Action queues
- • Revenue context
- • Notifications
Systems it reads
- • Google Sheets
- • HubSpot
- • Slack
The record model
What a headline metric has to carry.
- Metric definition
- Shown with the metric. Nearly every executive dashboard dispute is definitional and presents as a data quality problem, which sends the investigation in the wrong direction.
- Source system and reconciliation date
- When this last agreed with the system of record. A board number that has never been reconciled will be challenged at the worst possible moment.
- Drill-down path
- From every headline figure to the underlying records. A number you cannot open is a number that gets replaced by one somebody can explain.
- Normal band
- Derived from history, so a movement can be judged as unusual or ordinary rather than reacted to because it is visible.
- Recognition and currency rules
- Because revenue disagreements between systems are almost always timing, currency, or what counts as a customer rather than a data error.
- Access scope
- Per role, since an executive view aggregates data that is sensitive in combination even where each source is not.
- Definition change history
- So a trend that crosses a definition change is either valid or visibly invalid rather than quietly wrong.
How it runs
From a monthly pack to a standing view.
Step 01
Describe what a business dashboard has to do
Write the definition of each headline metric and get the leadership team to agree it. Nearly every executive dashboard dispute is definitional and presents as a data quality problem.
Step 02
Connect the systems of record
Billing, the CRM, and the operational systems each own part of the picture. Reading them directly is what allows a follow-up question to be answered rather than scheduled.
Step 03
Build the operating surface
Headline metrics against agreed definitions, trend views, and drill-down to the records — so a challenged number can be opened rather than defended.
Step 04
Start narrow
The five numbers the board actually asks about, defined and reconciled against their sources. Five defensible numbers beat forty charts, and the reconciliation is where the work is.
Step 05
Route the exceptions
A metric moving outside its normal band surfaces with the drill-down attached, so the first leadership conversation about it starts from evidence.
Step 06
Measure time from question to trustworthy answer
Measure the share of questions answerable in the meeting rather than deferred. That is what separates an operating view from a report that gets presented.
Implementation path
Building an executive view that survives scrutiny.
- 01
Write and circulate the definition of each headline metric before building. Expect at least one to turn out to mean different things to finance and to the operating team.
- 02
Baseline the assembly time for the current pack and how many questions per meeting become action items. Both make the case, and the second is the one that changes.
- 03
Reconcile each headline number against its source system before publication, and record the reconciliation. A board number that has never been reconciled will be challenged at the worst possible moment.
- 04
Keep drill-down on every figure. An executive view without drill-down produces confident numbers nobody can defend when questioned.
- 05
Build the narrowest useful version first: the three numbers the leadership team actually decides on, live rather than assembled.
- 06
Writing and circulating the definitions is a week of calendar time and mostly waiting for agreement; expect at least one metric to mean different things to finance and to the operating team. Reconciling five headline numbers against their sources is the technical core and is not optional if the numbers reach a board. Two to three weeks to a defensible first version.
- 07
Once the five numbers are defensible and drill-down works, add the normal band per metric so movement can be judged rather than merely observed. Segment breakdowns come after, and only for metrics somebody acts on.
Controls
Controls that matter.
Control 01
Each metric definition shown with the metric, so agreement is visible rather than assumed
Control 02
Drill-down from every headline figure to the underlying records, since a number that cannot be opened cannot be defended
Control 03
Access scoped by role, because an executive view aggregates data that is sensitive in combination even where each source is not
Examples
Three questions leadership stops waiting for.
The follow-up question
Drill-down turns what would be an action item into a two-minute answer, which changes the meeting from a status review into a decision-making session.
Three versions of revenue
Definitions attached to the metric convert a recurring dispute into a one-time agreement, and make it obvious when a system change alters what a number means.
The pack assembly
Two days of copying disappears, and the analyst’s time moves to the interpretation, which was the only part that ever required judgement.
How it goes wrong
Three ways executive views go wrong.
Forty metrics ship and leadership attention distributes evenly across things of very unequal importance.
Five to nine metrics that leadership actually acts on. Beyond that the view is a data catalogue, and a catalogue does not change a decision.
A dashboard figure contradicts a reported number and the dashboard loses the argument permanently.
Reconcile with finance before publication and record the reconciliation. The credibility cost of one contradicted number exceeds the effort of checking all of them.
Daily refresh is added to a monthly metric and leadership starts managing the noise.
Match refresh to the decision cycle and show the normal band. Currency matters far less than definition and lineage at this level, and it invites a kind of attention that is rarely productive.
Limitations and considerations
What an executive view cannot resolve.
- An executive dashboard cannot resolve a disagreement about strategy. It makes the disagreement more precise, which is progress and is often mistaken for resolution.
- Aggregated leadership views invite management by metric, including on figures that are proxies for what actually matters. The definition step is the only protection, and it is a judgement rather than a technical control.
- Live executive data changes meeting behaviour in ways that are not all good. Some organisations respond to daily visibility of a monthly metric with daily intervention in normal variation.
- If leadership is three people who already see the operational systems, this is ceremony. If the definitions cannot be agreed, the dashboard becomes the venue for the argument, and the argument is a management problem rather than a reporting one.
- Connector coverage varies: Google Sheets, HubSpot, Slack are representative rather than guaranteed, and the fields exposed depend on your workspace permissions.
FAQ
Build a business dashboard with AI: common questions.
How many metrics should a business dashboard have?
The number leadership genuinely acts on, which is almost always between five and nine. Beyond that the view becomes a data catalogue and attention distributes evenly across things of very unequal importance.
Why do our systems disagree about revenue?
Usually recognition timing, currency handling, or what counts as a customer, rather than a data error. Writing the definitions is what surfaces which of the three it is, and it is nearly always one of them.
Do we need finance to sign off?
For anything that reaches a board, yes. A dashboard number that contradicts a reported figure creates a problem out of proportion to its usefulness, and the reconciliation is cheap by comparison.
Should it be real time?
Rarely. Most leadership decisions operate on monthly and quarterly cycles, and daily refresh on a monthly metric mostly generates reaction to noise. Currency matters less than definition and lineage.
What should the first version contain?
The three numbers the leadership team actually decides on, live rather than assembled. Everything else waits until that one is genuinely used.
How will we know whether it worked?
Measure time from question to trustworthy answer against the baseline taken before anything changed.
Related pages
Keep exploring
Start with ARIA
Ask ARIA to build it.
Describe the website, application, workflow, or operating surface you need. ARIA plans, connects, builds, tests, and keeps refining it — inside the permissions you set.
- ARIA acts only through the systems and permissions you connect.
- Connections use scoped credentials you can change or revoke.
- Actions are recorded, and consequential ones can require approval.
Start here
Build a business dashboard around the process you actually run.
Agree the definitions first, reconcile each headline number to its source, and keep drill-down on every figure.