Every KPI in One Table. Nobody Decided Which Ones Come First.
[the following content is created with artificial intelligence technology]
I have seen this report more than once. Every KPI sits in one summary, with several ways to sort it, and the user selects what they want.
It is a good page. It should not be the first one.
KPI overload is usually told as a story about too many numbers, and the advice is to cut them. That advice throws away what this page gets right. A complete, sortable list answers questions nobody predicted. A controller chasing a variance goes straight to it. An analyst sorts by the worst performer and starts there.
Now open the same report with ten minutes before a meeting. You have every KPI in front of you and no signal about which ones matter this week. You need to know what you are looking for before the report shows you anything.
The page has moved a decision. The people who commissioned the report never ranked the KPIs, so each person who opens it does that alone.
If you build these reports, you know how the list got there. Each department asked for its number, and each request was reasonable. Nobody in the room wanted to tell a director that their KPI belonged further back.
A sortable table settles that tension politely. Everything is included and nothing is ranked. It is the same move that produced 40 reports and zero decisions, on a single page.
Keep the table. Put 2 or 3 pages in front of it.
Those pages carry the KPIs the business has agreed to read first. They state the position and show where it needs attention. Someone with ten minutes gets what they came for and leaves.
The KPIs themselves tell you how to arrange those pages, because they are rarely independent. Most belong to a family, such as delivery, margin or quality. Inside a family, some KPIs are consequences and others are causes.
Some links between them are arithmetic. OEE is availability times performance times quality, so a fall in OEE always traces back to one of the three.
Others are claims. A stopped line is a likely cause of a late delivery, and buffer stock or spare capacity absorbs some stops. A flat table hides that difference, because it shows every KPI as an equal row.
So lead with the consequences, the results the business answers for, and place their causes beside them or one page on. The reader sees what moved, then why.
Arithmetic links have to add up on the page. Cause-and-effect links stay labelled as claims. Kaplan and Norton said as much about their own scorecard: "A strategy is a set of hypotheses about cause and effect." Check those links on your own data, and check them again once the KPI becomes a target.
The reader who wants more keeps going, and reaches the full table with a question in mind, which is what that table is good at.
We structure our own reports this way: Overview, then Exploration, then Detailed.
The pages are the easy part. Choosing what goes on them is the work, because some KPIs have to move back.
Moving a KPI back is a different conversation from removing it. People agree to have that one. Every KPI stays in the report. The director whose number sits on the Exploration page still has it, one click further on.
We have all shipped the complete list because complete felt safe. It draws no objections on day one. The reaction comes later: the report is too complicated, and people go back to their Excel file.
Take the KPI list of the report you know best. Ask two things about each line. Which decision changes when this number moves? Who makes that decision?
These are close to the questions that test a real-time request. A KPI with a clear answer to both is a candidate for the first pages. The rest stay in the table, where the people who need them still find them. Then group the first-page candidates by family, and mark which ones drive which.
When two people give different answers about the same KPI, keep that line. Their disagreement is the conversation the list skipped.
If you fund BI projects, the order is yours to ask for. A report that shows everything leaves every reader to decide what matters. Agree which KPIs come first before the build starts, and write that order into the specification. The complete list stays, one page further back.
A sortable table of every KPI is a sound page, and the wrong first page. Each reader who lands on it has to rebuild priorities the business never set. One structured session settles which KPIs come first and what moves to Exploration. Nothing gets deleted. → Agree the order before you build
Agree the order before you build.
One structured session settles which KPIs come first and what moves to Exploration. Nothing gets deleted.
Book a 30-Minute Discovery CallFrequently asked questions
How many KPIs should a Power BI report page show?
No fixed number holds up, so start from the reader instead. A page carries the KPIs one reader needs for one recurring decision. When two audiences need different numbers, give them different pages.
Should I delete the KPIs nobody uses?
Not on usage data alone. The usage metrics report in Power BI counts page views, so a single table page tells you nothing about which of its KPIs anyone read. A KPI read only at the quarterly close still belongs. Ask the decision owners, and move a KPI back before you remove it.
What belongs on the first pages, and what belongs on an exploration page?
The first pages answer where we stand and what needs attention, with no input from the reader. An exploration page answers a question the reader already holds. A quick test: if the page only makes sense after someone picks a sort order, it is an exploration page.
How do you tell a director their KPI is not on the first page?
Separate the two statements: the KPI stays in the report, and only its position changes. Then ask which decision it drives and how often. In many ASEAN organisations this goes better one to one, before anyone meets, so nobody gives up a KPI in front of peers.