A KPI dashboard is not a collection of charts. It is an agreement about which numbers matter, how they are defined and who is allowed to act on them. Most failed dashboard projects start with a tool demo and only later discover that sales, finance and operations calculate revenue differently.
This buying guide is for an SME that wants a dashboard people actually use. It covers the requirements to settle before comparing vendors, the questions that reveal whether a platform can handle your data, and the commercial terms that prevent a cheap pilot from becoming an expensive annual subscription.
Start with the decisions, not the charts
List the five to seven decisions the dashboard should support. Examples: which products are losing margin, which branches are behind target, which customers are overdue, which support queue is breaching service levels, which marketing channel produces qualified leads. A chart that does not change a decision is decoration.
For each decision, define the metric, the owner, the source, the refresh frequency and the action. If nobody owns the metric, the dashboard will become a reporting museum.
Define the metric layer before the vendor
Ask each function how it calculates the number. Revenue may be gross or net of returns; active customer may mean logged in, ordered or paid; overdue may be based on invoice date or due date. Write the definition, the formula and the exclusions. This is the semantic layer, even if it lives in a spreadsheet at first.
Kaplan and Norton's balanced scorecard work is still useful here: financial measures alone are lagging indicators. Add customer, process and learning measures where they change a decision. Do not add twenty KPIs because the tool can display them.
Data sources and refresh
- Systems of record: accounting, ERP, ecommerce, CRM, helpdesk, POS, payment gateway.
- Spreadsheets: useful for planning, dangerous as a production source unless owned.
- External data: ad platforms, courier, marketplaces, bank feeds.
- Refresh: real time for operations, hourly for inventory, daily for finance, monthly for strategy.
Ask the vendor how it connects: native connector, API, scheduled file or manual upload. Ask what happens when the source schema changes. A dashboard that breaks silently is worse than no dashboard.
Visualization and access
The best chart is the one that answers the question with the least reading. A trend line for direction, a bar chart for comparison, a table for detail, a single number for a target. Avoid pie charts with many slices and dual axes that imply a relationship that does not exist.
Access matters as much as design. A branch manager should see their branch, not every branch. A salesperson should not see salary data. Role-based access, row-level security and a clear data owner are requirements, not nice-to-haves. If the dashboard contains personal data, the Personal Data Protection Act, 2026 applies.
Vendor scorecard
- Data connectors: the systems you actually use, not a logo wall.
- Semantic layer: reusable definitions, not a different formula in every chart.
- Access control: role and row-level security, SSO and audit logs.
- Refresh and reliability: schedule, failure alerts and historical backfill.
- Mobile experience: a usable phone view, not a shrunken desktop.
- Export and ownership: can you export the data, definitions and queries?
- Support: named contact, response time and implementation help.
- Pricing: per user, per viewer, per data volume or per connector; overage terms.
- Exit: what happens to your dashboards and data if you leave?
Pricing questions
Dashboard pricing is often per user or per capacity. A low viewer price can hide a high implementation or data-warehouse cost. Ask for a three-year total cost that includes connectors, storage, refresh, implementation, training and support. Ask what happens when usage doubles. Ask whether a viewer who only opens a weekly report needs a full licence.
The pilot
Pick one decision and one data source. Build a dashboard that answers one question, with a defined metric and an owner. Run it for two weeks and compare the number with the existing report. If the dashboard cannot reconcile to the current number, fix the definition before adding more charts.
A one-page requirements template
Before you compare vendors, fill one page with these rows. It will save weeks of demos.
- Decision: what action the dashboard should change.
- KPI: the metric name, formula, target and owner.
- Source: the system of record, the field and the refresh frequency.
- Access: who can see all rows, who sees only their branch or team.
- Action: what the viewer does when the number is off track.
- History: how far back the data must go and how corrections are handled.
- Export: whether the data and definitions can leave the platform.
Common buying mistakes
- Buying a visualization tool before defining the metrics.
- Letting each department calculate its own version of the same KPI.
- Choosing a platform on chart variety instead of data reliability.
- Ignoring row-level access until after the first sensitive report.
- Underestimating implementation, data cleaning and training.
- Signing a multi-year contract before a reconciled pilot.
Where this fits in our work
If you want help defining metrics and building the reporting layer, start with Operations Engine. The pricing page explains scope. For the architecture, read how a KPI dashboard works.
Sources
- Kaplan and Norton, "The Balanced Scorecard: Measures That Drive Performance", Harvard Business Review, 1992.
- Microsoft, Power BI guidance, accessed 16 November 2026.
- Google Cloud, Looker documentation, accessed 16 November 2026.
- Data-to-Viz, chart selection guide, accessed 16 November 2026.
- DataGuidance, "Bangladesh: Personal Data Protection Act 2026", accessed 16 November 2026.
Responses (0)
No comments yet. Add the first one below.
Leave a response
Your email address will not be published. Required fields are marked *.