I have built more than a dozen dashboards over the years, and I can tell you the uncomfortable truth: most of them stopped being opened within a month. Not because the charts were ugly. Because they were decoration, not decision tools.
A business dashboard earns its existence in one way only: it changes what you do on Monday morning. If you look at it and nothing about your week changes, it is a screensaver with your logo on it.
This article is about how to build the other kind, the one your team actually checks. The method is simple, and it starts nowhere near the software.
The 30-Chart Problem
Here is the typical failure pattern. A business owner asks for a dashboard. The vendor, wanting to look thorough, asks "what data do you have?" and then charts all of it. Sales by month, sales by product, sales by region, website visits, follower counts, average order value, twelve more.
The result looks impressive in the demo. Thirty charts, six colors, a filter panel. Three weeks later nobody opens it, because thirty charts answer no specific question. When everything is highlighted, nothing is.
I saw this at a distribution company in Tangerang. They had paid around Rp 60 million for a BI implementation with over forty visualizations. When I asked the operations manager which chart she used, she opened Excel instead. The dashboard was for the boardroom TV. The spreadsheet was for work.
Start From Decisions, Not Data
The fix is to reverse the direction. Do not start with "what data do we have?" Start with "what decisions do we make every week?"
Sit down with the two or three people who actually run the business and list the recurring decisions:
- Do we reorder stock this week, and how much?
- Do we chase overdue invoices, and which ones first?
- Do we increase the ad budget or pause it?
- Which branch or salesperson needs a conversation?
- Do we have enough cash to cover the next payroll?
Every one of those decisions is informed by a small number of figures. Reorder decisions need stock cover in days and sales velocity. Invoice chasing needs an aged receivables list. Ad spend needs cost per qualified lead, not impressions.
Work backwards from each decision to the one or two numbers that actually inform it. In my experience you end up with five to eight numbers total. That is your dashboard. Everything else is optional context at best.
The Five-Number Test
If you want a rough filter, every number on a business dashboard should pass three questions:
- Does someone act on it? Name the person and the action. "If receivables over 60 days exceed Rp 200 million, Dina calls the top five accounts." If you cannot write that sentence, cut the chart.
- Does it move weekly? Numbers that barely change month to month belong in a monthly report, not a daily screen.
- Is it a cause or a trophy? Revenue is a trophy, it already happened. Quotes sent, conversion rate, and stockouts are causes. Dashboards that lean on causes let you steer. Dashboards full of trophies let you watch.
Vanity metrics fail all three. Total followers, cumulative downloads, lifetime revenue: they only go up, they demand nothing, and they exist to make the room feel good. Cut them without mercy.
Automate Only What Survived the Cut
Only after you have your short list should you talk about tools, and this is where owners overspend. You do not need an enterprise BI platform to display six numbers.
For most Indonesian SMEs in 2022, the pragmatic stack looks like this:
| Situation | Sensible tool | Realistic cost |
|---|---|---|
| Data lives in spreadsheets | Google Data Studio on top of Sheets | Free plus setup time |
| Data lives in one system (POS, accounting) | The system's built-in reports, scheduled by email | Usually already paid for |
| Data lives in several systems | A small custom dashboard pulling from each API | Rp 15 to 50 million one-off |
Notice the order. The expensive custom option is the last resort, justified only when the numbers you need genuinely live in multiple systems that do not talk to each other. I wrote more about escaping manual reporting in From Spreadsheet Reports to Dashboards People Check, but the core rule is the same: automate the five numbers that survived the cut, not the forty that did not.
One more practical point: freshness matters more than beauty. A plain table updated automatically every morning beats a gorgeous chart someone updates manually every other Friday. Manual updates die the first busy week, and a stale dashboard is worse than none, because people quietly stop trusting it.
Give Every Number an Owner and a Threshold
A dashboard nobody is responsible for is a dashboard nobody checks. Two habits keep it alive:
- Assign each number to a person. Not a department, a name. That person explains the number's movement in the weekly meeting, in two sentences or less.
- Set a threshold that triggers action. "Stock cover below 14 days" or "conversion below 3 percent" should mean something specific happens. The threshold turns a chart into an alarm.
Then anchor the dashboard to an existing ritual. If you have a Monday operations meeting, the dashboard is the agenda. Opening it becomes part of running the business rather than an extra chore. This link to a real decision-making moment is, honestly, the difference between the dashboards that survive and the ones that decay.
This is also where a business dashboard connects to strategy. If your five numbers do not map to what you claim your priorities are, one of the two is lying. I have argued elsewhere that your business needs a technology strategy, not just a website, and a decision-driven dashboard is one of the cheapest ways to make that strategy concrete.
The Practical Takeaway
Before you commission or renovate any business dashboard, do this exercise. It takes one afternoon and costs nothing:
- List the recurring weekly decisions in your business. Aim for five to ten.
- For each decision, write the one or two numbers that inform it.
- Kill every number that has no owner, no threshold, and no action attached.
- Automate only what remains, using the cheapest tool that refreshes daily.
- Bolt the result onto a meeting that already exists.
If your current dashboard would lose 80 percent of its charts under this test, that is not a failure of the exercise. That is the exercise working. Five numbers people act on will outperform fifty numbers people admire, every single week.