In this guide 6 sections
Use a BI tool like Power BI or Metabase when staff need internal charts, filters and scheduled reports. Build a custom dashboard when the audience is customers, when the owner needs a fast phone screen, or when each user must see only their own rows.
A BI tool is enough more often than people think
Power BI, Metabase and Looker Studio connect to a database or spreadsheet in an afternoon. They give you charts, drill-down, filters, scheduled emails and access roles without a development team. Each one reads PostgreSQL, MySQL, Excel files and Google Sheets directly, so your data rarely moves before the first chart appears.
For finance, sales and operations managers who sit at a desk and read a dozen standard views, that is usually the right answer. You pay a licence or hosting cost. You maintain no code.
Start there unless one of these cases applies. Metabase on a PostgreSQL read replica costs little to try. Custom work earns its cost only where a tool fights you. Count the people who will edit reports as well as view them, because a tool your own staff can change beats a prettier screen they cannot.
- Internal users on laptops
- Standard chart types and tables
- Reports that refresh daily or hourly
- Roles that map to departments
- A small team that will edit reports itself
When the owner’s phone is the main screen
Owners often want five numbers that open in a second: today’s sales, cash in bank, orders pending, receivables over thirty days, and one alert tile. BI tools handle mobile layouts, but they load slowly and ask for a login each time. These are single numbers. A plain page that loads at once serves them better than a report canvas.
A custom screen can cache the figures, open straight from a home-screen icon, and show a large red number when stock or cash crosses a line you set. It can also send a push message. Add a threshold alert, such as a fast-moving item falling below its reorder level, and the screen becomes something the owner acts on rather than glances at.
If the owner will glance at it twenty times a day, a purpose-built view matters more than any chart type.
When customers or partners need to see data
Outsiders are where tools strain. A distributor wants dealers to see their own orders and statement. A contractor wants clients to see progress by site. Embedding a BI report for them often means per-viewer licences, branding limits and a clumsy login. Dealers and clients also expect statements, delivery proofs and downloadable invoices next to the charts. That turns a dashboard into a small application with its own support needs.
A custom portal can use your own sign-in, carry your logo and fonts, and present only a few clear views. It can also take action, such as paying an invoice or raising a query. Single sign-on with their own email, a branded domain and a clear log of who opened what are features customers notice immediately.
Check the licence terms first. External sharing is where costs and rules differ most.
- Dealer or client portals with their own login
- White-labelled screens in your brand
- Actions beside data, such as approvals
- Pricing that does not scale per external viewer
When each user must see only their own rows
Row-level security exists in the major tools, and it works. Setup is delicate. One wrong filter leaks data. Territory managers, franchise owners and clinic heads who must never see each other’s figures raise the stakes. The danger is usually a rule that grows over time, such as a franchise owner who later gains a second territory, and breaks a filter written for one.
Some rules go beyond a region field. Take a user who sees three branches plus one product line. Those rules are easier to enforce in the application layer, with tests and an audit log. Write a test that logs in as each role and checks the row counts against known answers before every release, so no one finds a leak by accident.
Ask who would be harmed by a leak. If the answer is a customer or a regulator, spend on control.
Choose by the audience, then test with a prototype
Write down the audience, device, data limits and refresh needs for each dashboard. Most land in the BI column. One or two may need building. Include a column for how many people will view each dashboard, as licence costs and server load both scale with that number.
Prototype that one in a tool first. Then watch where users stall. Those limits become the exact list of what the custom build must do.
Mixing is fine. Many firms run Metabase for managers and a small custom view for the owner and the dealers. Both read the same warehouse tables, so every total agrees.
Build or buy scorecard
A scorecard with one row per dashboard. Columns cover audience, device, number of viewers, external or internal, data limits per user, refresh speed, actions needed, and the verdict of BI tool, custom, or both. A notes block lists licence terms to check with each vendor.
Open a blank worksheet to printQuestions owners ask
Is Power BI or Metabase enough for a small business?
Often yes. For internal reports on a desk, either gives filters, charts and scheduled emails without writing code. Move to custom work only when a clear need appears.
When is a custom dashboard worth building?
When customers need their own login and views, when the owner needs a fast phone screen, or when per-user data limits are too complex for a tool’s settings.
Can a custom dashboard sit beside a BI tool?
Yes. Many companies keep a BI tool for managers and add a small custom screen for the owner or for external partners, both reading from the same data.
Do BI tools support row-level security?
The major ones do. It needs careful setup and testing, because a wrong rule shows someone data meant for another person.