Dashboards fail in a very particular way: beautiful at sign-off, opened a few times in the first week, then never again. The cause is almost always the same — it answers questions nobody asked, and does not answer the one question a director actually has each morning: is there anything unusual today that needs me?
Eight numbers, enough to run on
| Indicator | Answers | How often |
|---|---|---|
| Today’s revenue against the same period | Are we above or below the normal rhythm? | Daily |
| Gross margin by category | We are selling, but are we earning? | Weekly |
| Cash balance and 30-day cash forecast | Will we be tight this month? | Weekly |
| Overdue as a share of receivables | Is our money sitting with customers? | Weekly |
| Value of stock older than 90 days | How much money is dead on the shelves? | Monthly |
| Orders past their delivery date | Which customer is about to complain? | Daily |
| Revenue by salesperson or branch | Who is pulling, who is slipping? | Weekly |
| Repeat-customer rate | Is growth coming from new or existing customers? | Monthly |
These eight cover the four faces of a business: are we selling, are we earning, do we have cash, are we keeping customers. A ninth should only be added when you can name the specific decision it serves.
More important than the number: the threshold
A number alone says nothing. Is "revenue today: 42 million" good or bad? It only means something beside a comparison point and a threshold.
For each indicator, define three bands: normal, watch, act now. And state the action for the third — for example, "if overdue exceeds 15% of receivables, stop deliveries to customers more than 60 days late".
A good dashboard is not a place to look at numbers. It is what tells you when to leave it and go deal with something.
What to remove from the main screen
- Pie charts with more than five slices. People cannot compare slice areas; use a ranked bar chart instead.
- Decorative gauges and progress bars. Large footprint, little information.
- Year-to-date totals in the best position. They barely move day to day, so they help no daily decision.
- Detail tables of hundreds of rows. Detail belongs on a sub-page, not the overview.
- Indicators nobody owns. If a number worsens and nobody has to do anything, it does not belong on the dashboard.
Three comparison points every indicator needs
- 1Against the same period last year — removes seasonality, essential in trades with Tết and clear peak seasons.
- 2Against the previous period — shows the short-term trend.
- 3Against plan — shows the gap to target.
Missing comparison points is the most common reason a dashboard "looks good but is unusable".
Who sees what
A dashboard should have at least three versions: management (everything, including cost prices and gross margin), branch or department managers (full operational data for their own area, no cost prices), and staff (only their own targets and results).
One shared screen for everyone always fails at one end or the other: either it exposes sensitive data, or it is so empty nobody uses it.
The prerequisite
Every number above is only trustworthy when the data beneath it is clean: stock matching reality, receipts matched to invoices, no duplicate products. Building the dashboard before cleaning the data is the surest way to produce a handsome tool the whole company quietly understands to be "not really accurate".
Want to talk specifics?
SealCore surveys at your premises and sends a fixed quote after the first session — including when the conclusion is that you do not need custom software.
