Key takeaways
- Start with the goal and the audience: which decisions does the dashboard need to support?
- Choose a handful of KPIs per audience and record each KPI's definition, source, owner and target.
- Sketch the dashboard first with fictional figures; making changes on paper is cheaper than in a built model.
- Show every KPI with context, such as a target, last year or a trend.
- Arrange access per role and plan maintenance: a dashboard without an owner quickly goes out of date.
What a KPI dashboard is, and what it is not
KPI stands for key performance indicator: a measurable value that shows whether you are on track to reach a goal. A KPI dashboard brings those values together on one screen, with a comparison against a target and a trend. A management dashboard is a KPI dashboard for the board or management team, with the figures at the level of the whole organisation.
A KPI dashboard is not a report containing every available figure. The aim is to see at a glance where action is needed. The details are on detail pages or in operational reports.
Step by step: how to build a KPI dashboard
Building a KPI dashboard works best in a fixed order. The first five steps happen before anything is built:
- Set the goal and the audience: who looks at it, how often, and what decisions do they make?
- Choose the KPIs that belong to those decisions, and no more.
- Define each KPI: calculation, source, filters, owner and target.
- Map the sources and check whether the data is available and reliable.
- Sketch the dashboard first, with fictional figures, and discuss the sketch with the users.
- Build the dashboard according to fixed design rules.
- Arrange access per role, so that everyone sees the right figures.
- Plan maintenance: refreshes, checks on the definitions and periodic evaluation.
Choosing and defining KPIs
A good KPI can be influenced, can be measured from available data and has an owner and a target. Combine lagging indicators, such as revenue and margin, with leading indicators, such as pipeline value or order intake, which show earlier where things are heading. Avoid figures that mainly look good but never lead to a decision.
The definition is at least as important as the choice. "Revenue" can be invoiced or ordered, including or excluding VAT, with or without credit notes. So for each KPI, record:
- Name and short description.
- Calculation, including filters and exceptions.
- Source, table and field.
- Owner: who is responsible for the definition and the follow-up.
- Target, and when the KPI turns red.
- How often the KPI is refreshed.
Examples of KPIs by department
Which KPIs fit differs per organisation. These examples are a useful starting point when choosing:
| Department | Example KPIs | Typical source |
|---|---|---|
| Board | Revenue against budget, gross margin, EBITDA, liquidity | ERP or accounting system |
| Finance | Days sales outstanding (DSO), open items, costs against budget, cash flow | ERP or accounting system |
| Sales | Pipeline value, conversion rate, average order value, new customers | CRM |
| Marketing | Leads per channel, cost per lead, website conversion | Marketing tools, CRM, web analytics |
| Operations and logistics | On-time delivery, lead time, inventory turnover, order picking error rate | ERP, warehouse management system |
| HR | Absenteeism rate, staff turnover, open vacancies, staffing against planning | HR system, planning |
| Customer service | Response time, resolution time, customer satisfaction (NPS or CSAT) | Ticketing system, surveys |
| Retail | Revenue per hour, in-store conversion, average transaction value, staff costs as a percentage of revenue | Point-of-sale system, staff planning |
Sketch the dashboard before you build
A step that is often skipped is sketching. Make a sketch or clickable prototype with fictional figures and walk through it with the people who will use the dashboard. For each tile, ask: what do you do if this number is red? If nobody can answer that question, the tile probably does not belong there.
At Data Analytics, the company behind ENABLE, this is the standard way of working: first visualise the end product, then connect and model the data, and only then set up the portal. That way, everyone knows in advance what will be built, and you avoid having to rebuild a data model afterwards.
Design rules for a management dashboard
A management dashboard is often viewed briefly and under time pressure. These design rules help to make it quick to read:
- Put the most important item top left; that is where the eye starts.
- Limit the number of visuals per page and use detail pages for the specifics.
- Give every number context: a target, last year or the previous period. Revenue on its own says little.
- Use colour only for meaning, such as status against target, and combine red and green with an icon or text for colour-blind readers.
- Match the chart to the question: a line for a trend, bars for a comparison. Avoid 3D and pie charts with many segments.
- Write titles that say what you see, such as "Revenue per month against budget".
- Show when the data was last refreshed.
- Check how it looks on mobile if managers view the dashboard on the go.
Role-based access and maintenance
Not everyone needs to see everything. The board sees the whole organisation, a regional manager their own region and an external partner only their own figures. You handle this with roles and row-level security, so that you maintain one dashboard instead of ten copies.
Then maintenance begins. Make sure every KPI has an owner, monitor whether refreshes succeed and review each quarter whether the KPIs still fit the goals. Dashboards that nobody opens any more are better removed.
In ENABLE, you manage access with access profiles and row-level security, everyone sees when the data was last refreshed, and Data Analytics uses the daily Power BI health check to track which dashboards are opened and whether they are healthy. You record monthly targets in the built-in Targets app. With the Dashboard Builder, available as a pilot, you create your own KPI tiles with a comparison against last year.
Common mistakes with a KPI dashboard
A KPI dashboard that is rarely used often stumbles over one of these points:
- Too many KPIs, so that nothing stands out any more.
- No agreed definitions, so that every department quotes a different number.
- Building before the users have seen the sketch.
- Manual Excel files as a source, which break as soon as someone moves a column.
- No target, so that nobody knows whether a number is good or bad.
- No owner, so that after a few months the dashboard is no longer correct.