Data & Analytics
Power BI for Business Analytics
Models, measures, and reports that answer a business question.
Power BI for Business Analytics focuses on the report a team will actually open. You will shape data, write measures, and design a page around a question. Excel comfort helps. The course does not assume you have built a model before.
Course syllabus
01. The question
- What the report is for
A decision or a check, not a gallery of charts.
- Write the question in one line
The sentence the report is allowed to answer.
- Who will open this page
A manager, a shop lead, or you. The page changes with the person.
- The number they already trust
Start from a figure they recognise, then add the split.
- What a chart will not decide
A picture is not a recommendation. Say the limit out loud.
- The grain of the data
One row is one order, one day, or one store. Pick it and keep it.
- A question you will not answer
If the export cannot support it, it does not go on the page.
- The page title is the question
Not 'Dashboard'. The thing they came to check.
02. Tables
- Tables Power BI can trust
Keys, types, and the mess to fix before you model.
- One header row
Names in the first row, and no title sitting above them.
- A key that identifies the row
The column that is unique, or the pair of columns that is.
- Dates as dates
A column Excel or Power BI can sort, not text that only looks like a date.
- Numbers stored as numbers
A currency typed as text will not sum.
- Blanks and what they mean
Missing is not zero unless the business says it is zero.
- Split a column that hides two facts
City and state in one cell will fight you later.
- A table named for its grain
Orders, not 'Sheet1'.
03. Load
- Get data and look at the preview
The first twenty rows, and whether the types look right.
- Rename a column in the query
The name the report will show, chosen before the visual.
- Remove a column you will not use
Less to break, and a model that is easier to read.
- Filter a row that is not data
A totals row at the bottom of an export does not belong in the table.
- A step you can explain
Each applied step should have a reason you can say.
- Refresh and see the same shape
The query should survive a new export of the same kind.
- Close and apply
The model sees the result of the steps, not the original file.
- Check the row count
If the count changed, know why before you build a visual.
04. Relationships
- Relationships
How tables connect, and what a wrong link does to a total.
- The one side and the many side
One customer, many orders. The arrow follows that fact.
- A relationship on the key
The columns that mean the same thing, not two columns that happen to look alike.
- Why a total doubled
A many-to-many or a wrong key repeats rows.
- A date table
One row per day, so time intelligence has somewhere to stand.
- Do not join on a name
Two people can share a name. An id does not.
- Inactive until you need it
A second path between tables is a choice, not an accident.
- Draw the model in words
Orders to customers, orders to products, orders to dates.
05. Measures
- DAX you will reuse
A few measures, named so the report stays readable.
- A measure is not a column
The total is calculated. It is not stored on each row.
- Sum of a column you trust
The first measure, named for the thing it counts.
- A count of rows
How many orders, written as a measure you can drop on a card.
- Filter a measure
This year's sales, as its own name, not a visual filter you forget.
- Divide only when the bottom is not zero
A ratio that does not blow up on an empty slice.
- Name measures in plain language
Sales amount, not 'M1'.
- A measure used on two visuals
Change it once and both visuals follow.
06. More DAX
- CALCULATE in one sentence
This measure, under an extra filter you name.
- A year-to-date total
From the start of the year through the selected date.
- Compare to last year
The same measure, shifted, then a difference you label.
- A percent of the total
This slice divided by the whole, with the whole held still.
- Keep the formula short
If you cannot read it aloud, split it into two measures.
- A measure that is blank on purpose
Hide a result when the slice has no rows.
- Do not write DAX in a title
The visual title is for the reader, not for the formula.
- Test the measure on a known month
A number you can check against the source.
07. The page
- Filters, slicers, and a page
Let someone narrow the view without breaking the numbers.
- One slicer that matters
Region or month. Not every column.
- A card for the answer
The number, large, with the question beside it.
- A bar for the comparison
Categories sorted by the measure, not by alphabet, unless alphabet is the point.
- A line only if time is the axis
A line of unrelated categories is a bar chart in disguise.
- Labels a person can read
No legend that requires a tour.
- The filter affects what you claim
If a slicer is on, the title should say so.
- A second page only for a second question
Do not split one answer across two pages.
08. Review
- Click every slicer once
The number should change in a way you can explain.
- A slice with no data
The page should look empty on purpose, not broken.
- Match one figure to the source
The same month, the same total, in Excel or the export.
- Remove a visual that does not answer
Pretty and unused is still unused.
- Write two sentences under the title
What this page is, and what it is not.
- A colleague's question
The thing they would ask, and whether the page answers it.
- Save the file with a clear name
The question and the month, not 'final final'.
- What you would add next
One measure, named, not a new chart type.
Lecture videos stay with the course. Playback for enrolled students will open once payments are available. Video links are not published on this page.