Opening Power BI before the KPI is named is how a team gets forty pages and still no decision.
The tool will draw whatever you feed it. It will not tell you that “revenue,” “sales,” and “bookings” were three different questions wearing the same label.
Name the metric first. Then the dashboard has a job.
A KPI is a sentence, not a chart
A usable KPI has four parts, written down:
- Name — one phrase, used in every meeting
- Grain — one row equals what (invoice, day, employee-week)
- Owner — who is allowed to change the definition
- Rule — what is in, what is out (tax, cancellations, transfers)
If any of those is missing, two files will disagree and both will look professional.
What goes wrong when you skip this
Someone builds a report because leadership asked for “a dashboard.” Pages multiply. Slicers multiply. The weekly meeting still starts with: which number are we using?
Finance brought a close file. Ops brought a snapshot. The dashboard brought last week’s refresh. All three are labelled the same KPI. None of them is lying. They are answering different questions.
Power BI cannot fix a missing definition. It can only make the confusion faster and prettier.
Do this before Desktop opens
Write one page. Not a deck. One page:
- Who is in the room, and what they must decide
- The one to three questions the view must answer
- The grain of the fact
- The source that is allowed to be the number
- What “done” looks like — usually one to three pages, not forty
That page is the spec. The model and the visuals come after it. If the spec cannot be written, the data is not ready for a dashboard. The work is still the definition.
One number in the room
Leadership does not need another tool. Leadership needs one named metric, at one grain, from one source, every week.
Tool choice — Power BI, Tableau, Excel — comes after that sentence exists. Building first, defining in week six, is how projects stall.
A short dictionary beats a gallery of charts. When the number moves, people should be able to point to the sentence that explains what moved — not to a visual title someone typed.
When a dashboard is the right output
After the KPI is named, a weekly operating view can sit on it: the same pack, the same way, every week. That may be Power BI. It may be a one-page read: movement, driver, next step.
If the data is still a mess, the next layer is cleanup — Python, R, or a tighter source — not more pages.
PeekGraph’s data analysis work is that thinking layer: KPI design, source of truth, and the weekly view. The dashboard is an output, not the starting point. Details are on the data analysis consulting page.
Name the metric. Then the meeting can decide.