An executive dashboard is a single screen with the metrics a company actually makes decisions on: money, sales, customers and workload. Its job is to show a deviation from plan within a minute, not to explain it — the explanation lives in reports and drill-downs. That is why the screen carries only the numbers someone has both the authority and the means to act on.
Key takeaways
A dashboard is a permanent window that people open every day and that refreshes itself. A report answers one question for a closed period, and someone puts it together by hand for a specific request.
Mixing the two is expensive. A dashboard built like a report stops being opened within a month: the numbers go stale and nobody is tasked with refreshing them.
| Trait | Dashboard | Report |
|---|---|---|
| Period | Current: day, week, month to date | Closed: month, quarter, year |
| Refresh | Automatic, on a schedule | Manual, on request |
| Purpose | Spot the deviation | Explain the cause and propose a fix |
| Volume | A few metrics on one screen | Dozens of tables and breakdowns |
| How it is read | Every day, in a minute | Once a period, in detail |
| What follows | A question to the owner of the number | A decision and an action plan |
Hence the selection rule: a metric stays on screen only if the team's actions move it. A macro indicator such as an industry index is useless there — you cannot assign anything based on it. What to do once a deviation is found is covered in the article on turning analytics results into an action plan.
A working set comes from four groups — money, sales, customers and operations — with one or two metrics from each. Such a screen answers four questions: how much did we earn, what is coming next, are we losing customers, and is the team keeping up.
Revenue and margin show the result; cash received and receivables show the ability to pay the bills. Revenue sits next to the plan and the previous period: a number without a basis for comparison says nothing about whether things are going well.
From the pipeline the screen takes the value of open deals, the conversion to payment and the average order value; from the customer base, the number of new and returning customers. How the stages work and where deals get lost is covered in the article on the sales funnel in CRM.
Operational metrics depend on the business: lead time and overdue jobs in services, stock levels and turnover in retail, utilisation and defects in manufacturing. They earn their place because a deviation shows up here before it reaches the money.
| Group | Metric | Where it comes from |
|---|---|---|
| Money | Revenue: actual and plan | Accounting system or 1C |
| Money | Margin by line of business | Accounting system, cost of goods from purchasing |
| Money | Cash received and receivables | Bank, payment services, 1C |
| Sales | Value of open deals | CRM |
| Sales | Conversion from lead to payment | CRM plus the payment fact |
| Customers | New and returning customers | CRM or point of sale |
| Customers | Average order value | CRM, point of sale, billing |
| Operations | Overdue tasks and tickets | CRM, task tracker, service desk |
As many as a manager can read in a minute: usually six to nine on the main screen, with everything else in drill-downs opened with a question already in mind. Check every candidate against this list.
Dashboard data sits in four places: leads and deals in the CRM, revenue and cost in the accounting system, traffic and website enquiries in web analytics, and the payment fact in the bank and payment services. The dashboard itself recalculates nothing: it displays what a scheduled job has already collected in one place.
Data is pulled through APIs, and every source has a request rate ceiling. Build it into the refresh schedule, or the export will start failing at the worst moment — the morning of reporting day.
That last figure explains why a dashboard needs storage of its own: a year-on-year comparison in standard Google Analytics 4 is impossible unless the data was exported in advance. Web analytics and CRM systems have other boundaries too — a cap on goals, a limited history, a fixed set of fields — and they are checked before a metric reaches the screen.
Important One metric, one definition. If sales count revenue by deal date and accounting counts it by payment date, the screen will carry two different revenues, and the business conversation will turn into an argument about numbers. Definitions are written down before the build and kept next to the dashboard.
The tool follows from the number of sources and from who will maintain the pipeline. A spreadsheet is fine while there is one source; past two, you need a system that refreshes the data itself.
| Tool | When it fits | Cost |
|---|---|---|
| Excel or Google Sheets | One source, and you need to start this week | Included in the office subscription |
| Power BI | Many sources, calculations and access control required | Pro — $14 per user per month, Premium Per User — $24, billed yearly |
| Metabase | Data in your own database, dashboard on your own server | Open Source — free, cloud Starter plan — $100 per month |
| A data mart inside CRM or ERP | The numbers are needed where the team works every day | Part of the cost of customising the system |
Licences are counted by the number of people who will open the dashboard, not by the number who build it. If every branch manager has to see it, a free tool or your own data mart works out cheaper than a paid seat for each of them.
The order does not depend on the tool: decisions first, then metrics and definitions, then sources, and charts only at the end. A build that starts with choosing a BI system usually ends with a handsome screen nobody uses.
Refresh frequency equals decision frequency. Data arriving faster than you are prepared to react to it only adds noise.
Every metric needs an owner: the person who explains a deviation and answers for data quality. The second must-have is a change log — who changed a formula, when and why. Without it, in six months nobody will remember why last year's revenue on the dashboard does not match the revenue in the closed report.
Tip Start with one screen and three metrics you already check by hand. Once they match your manual calculation two periods in a row, add the next ones. A screen of twenty tiles built in one go rarely survives into its second month.
Dashboards fall out of use for six reasons, and none of them has to do with the choice of tool.
The main local feature is two currency circuits: part of the prices and settlements in dollars, accounting and reporting in soum. If the dashboard has no fixed rule for the exchange rate, the same revenue will differ between two reports.
We start with a conversation about decisions rather than charts: what you want to see each morning and what you will do if a number drops. Then we go through the sources — what is in the CRM, what is in 1C, what is in the payment services, which data is typed in by hand and where the figures diverge. The outcome is a list of metrics with definitions and an export design, and only then does the screen get built.
We have no separate service called a dashboard: the work sits inside a systems project. Putting data and reports in order in a working CRM starts at $2,000; connecting several systems and building reporting on the combined data starts at $4,000. We quote exactly after reviewing the sources: the price depends on how many there are and on how much data is still moved by hand.
Let’s discuss your project
Tell us what you need, and we will estimate the timeline and cost and suggest a solution.
Compare the definitions first: sales usually count revenue by deal date while accounting counts it by payment date. Pick one date for the dashboard, write it into the metric definition, and only then compare the figures.
Yes, while there is one source and few metrics: Excel or Google Sheets give you a working screen in a week. Once there are more than two sources, manual copying falls behind and you need a system that refreshes itself.
Count readers, not authors. A Power BI Pro seat costs $14 per user per month, so with many managers to serve, a free tool or Metabase Open Source on your own server works out cheaper.
Calculate the same metrics for a closed month by hand and compare. A discrepancy means an error in the export pipeline or in the definition, not a quirk of the tool; repeat the check two periods in a row before adding new metrics.
Choose a conversion rule and write it down: the rate on the deal date or on the payment date. Take the official rate from the Central Bank of Uzbekistan and pull it automatically rather than typing it in, or two versions of the report will diverge.
In storage of your own — a database or a data mart inside the ERP that scheduled exports feed. Standard Google Analytics 4 properties keep user and event data for no longer than fourteen months, and history deleted on the service side cannot be restored.
Each metric has its own owner, who explains deviations and answers for data quality, while the dashboard as a whole has a person who watches the exports and keeps the log of formula changes.
Sources
Cover photo: Anthony Borshon Gomes, Pexels
