Business dashboards get used when they answer a small number of questions the team already asks every week, from figures everyone trusts, on a screen people actually open. They get ignored when they show everything the software can chart. This guide explains how to choose the measures, set up one source of truth, decide how fresh the data needs to be and who should see what, and it lists the mistakes that most often turn a dashboard into a page nobody visits.
Start with questions, not charts
A dashboard is a single screen that shows the current state of a few important measures. The usual way to build one is to connect a tool to a database and start adding charts. The better way is to begin with the questions people ask in meetings and on the phone: are we getting paid on time, which jobs are running over, do we have enough stock for next week, how busy is the front desk.
For each question, write down who asks it, how often, and what they would do differently depending on the answer. If nobody would act on a number, it does not belong on the dashboard. A cafe group's owner who asks "which site is over on wages this week" will change a roster because of the answer.
Choose a few measures
Keep the first version to a handful of measures per audience. A screen that can be read at a glance is the goal. A useful measure has four properties:
- It is defined in writing. "Revenue" could mean invoiced, paid, including or excluding GST. Pick one and write it down.
- It has a comparison. A number alone says little. Show it against a target, the same period last year or the previous week.
- Someone owns it. One person is responsible for the number and for what happens when it moves.
- It can be traced. A reader can click through, or ask, and see the records behind it.
The table below shows how a question becomes a measure and where the data comes from. These are examples; yours will differ.
| Question | Measure | Data source |
|---|---|---|
| Are we getting paid on time? | Overdue receivables by age band, and average days to pay | Accounting system |
| Is the sales pipeline healthy? | Value of open quotes by stage, and quotes won against quotes sent | CRM |
| Which jobs are running over? | Hours and materials used against the quoted amount, per job | Job management system and timesheets |
| Will we run out of stock? | Days of cover for the fastest-moving items | Inventory or point-of-sale system |
| Are customers waiting too long? | Open support requests by age, and time to first reply | Help desk or shared inbox |
| Is each site covering its costs? | Sales, wages and gross margin by location | Point of sale, payroll and accounting |
One source of truth
Nothing damages a dashboard faster than two reports showing different totals for the same thing. A single source of truth means that each measure is calculated once, in one place, from agreed data, and every report reads that result.
In a small business the data usually lives in several products: accounting, a CRM, a point-of-sale system, spreadsheets. There are two common ways to bring it together. The simpler one is to let the dashboard tool connect to each product and combine them inside the report. That is fine for a first version. The sturdier one is to copy the data on a schedule into a central database, often called a data warehouse, clean it there and point every report at it.
BigQuery is one product used for that central store. Google describes BigQuery as a fully managed data platform with a serverless architecture, which means there are no servers for you to run, and it is queried with SQL, the standard language for databases. A small business with modest data may not need a warehouse at all to begin with.
Whichever route you take, the unglamorous work matters most: matching customer records between systems, agreeing which system owns each type of record, and fixing bad data where it is entered. This is where business systems integrations and dashboards meet: if the CRM and the accounting system already agree on who the customers are, the dashboard is far easier to trust.
How often should the data refresh
Match the refresh to the decision. A monthly board pack does not need live data. A dispatch screen in a warehouse does. Decide measure by measure.
Tools also set their own limits. Microsoft's documentation on data refresh in Power BI says a scheduled refresh can be set for up to eight daily time slots when the data sits on shared capacity, or 48 time slots on Power BI Premium. The same page explains that if Power BI cannot reach a data source over a direct network connection, such as a database on a server in your office, a gateway has to be set up before refreshes can run. Check points like these before promising the team an hourly update.
Always show the time of the last refresh on the dashboard. People forgive data that is a few hours old. They do not forgive data that looks current and is not.
Who sees what
Decide access before you build. A dashboard often combines information that different people should see at different levels: an owner sees margin by site, a site manager sees only their own site, and wages detail stays with payroll. Write a short table of roles and what each may see, then use the tool's access controls to enforce it.
Give access by role, remove it when someone changes job or leaves, and be careful with sharing by public link, since a link that needs no sign-in can be forwarded to anyone.
Tools in general terms
The tool matters less than the questions and the data. The common types are:
- Reports built into your existing software. Accounting, CRM and point-of-sale products all include reports. If one system holds everything a question needs, start there.
- Spreadsheets. Familiar and adequate for a first prototype, but fragile once several people edit them.
- Dedicated dashboard tools. Microsoft describes Power BI as its business analytics platform, with Power BI Desktop for data modelling and report creation and the Power BI service for sharing and collaboration. Google's equivalent is Data Studio, which its documentation notes was previously called Looker Studio; it is a tool for building dashboards and reports with a drag and drop editor and for sharing them with a team.
- A custom dashboard. A screen built into your own software or website, suited to cases where staff or customers should see figures inside a system they already use.
Choose according to where your data already lives, which accounts your staff already sign in with, and who will maintain the reports after launch.
Common mistakes
- Too many measures. If everything is highlighted, nothing is. Remove charts that nobody has acted on.
- No written definitions. Two people argue about a number because each assumed a different meaning.
- Manual steps. A dashboard that relies on someone exporting a file every Monday stops updating the first time that person is on leave.
- Fixing data in the report. Corrections made in the dashboard layer hide problems that should be fixed in the source system.
- No owner. Without someone responsible, broken connections and stale figures go unnoticed.
- Launching and leaving. Businesses change. Review the dashboard with its users regularly and retire what is no longer used.
A sensible way to begin
Pick one audience and three or four questions. Build a first version from the systems you already have, check every figure against a report you trust, and use it in a real meeting.
Comingwave is an Australian technology company that builds data dashboards for small and medium businesses, usually as part of wider digital transformation work to replace manual reporting. If you would like help turning your questions into a working dashboard, get in touch for a quote.
Key takeaways
- Begin with the questions your team asks, and keep only measures someone will act on.
- Define each measure in writing and calculate it once, in one place.
- Set the refresh frequency by the decision it supports, and show the last refresh time.
- Decide who sees what before building, and manage access by role.
- Give the dashboard an owner and review it with its users.
Frequently asked questions
What should a small business dashboard show?
The few measures that drive decisions in your business, typically cash and receivables, sales or pipeline, work in progress, and one or two measures of service or stock. Each should have a written definition, a comparison such as a target, and a person responsible for it.
Do we need a data warehouse to build a dashboard?
Not to begin with. Many first dashboards connect straight to the accounting system, CRM or a spreadsheet. A central database becomes worthwhile when several systems must be combined, when reports are slow, or when different reports start to disagree.
How often should a dashboard update?
As often as the decision needs. Operational screens may need updates through the day, while financial summaries are usually fine daily or weekly. Check what your tool supports; Microsoft, for example, documents limits on the number of scheduled refreshes per day in Power BI.
Which dashboard tool is right for a small business?
The one that connects to your existing systems, works with the accounts your staff already use, and can be maintained by someone you can call on. Start with the reports built into your current software and move to a dedicated tool when you need to combine sources.