Prepare a small, reconciled dataset

In a new training report, choose Home → Enter data. Create the columns below and name the table Sales. Enter amounts as numbers without currency symbols. Set Channel to Text and both amount columns to Fixed decimal number. Each row represents one channel for one common reporting period; this is a teaching summary, not an invoice ledger.

For a production model, use the existing validated sales fact table and channel dimension. Agree treatment of VAT, credit notes, rebates and unallocated costs with Finance. Do not append this training summary to transaction data: doing so would double-count sales.

Create measures that keep their meaning

Select the Sales table and use New measure for each definition below. Create four separate measures, one block at a time. Keep their home table as Sales so the parameter references match.

Sales Value = SUM('Sales'[SalesAmount])

Cost Value = SUM('Sales'[CostAmount])

Gross Profit = [Sales Value] - [Cost Value]

Gross Margin = DIVIDE([Gross Profit], [Sales Value])

Format Sales Value, Cost Value and Gross Profit as GBP currency; format Gross Margin as a percentage with one decimal place. Do not convert these measures to text with FORMAT. DIVIDE returns blank for a zero denominator when no alternative is supplied; a blank margin is different from a genuine zero margin. Microsoft: DIVIDE.

The total margin is total gross profit divided by total sales. Averaging the three channel percentages would incorrectly give 33.9%, instead of 25.0%. This distinction matters whenever channels differ in size.

Create the field parameter

Choose Modelling → New parameter → Fields. Name it Metric, select Sales Value, Gross Profit and Gross Margin, and select the option to add a slicer to the page. Create the parameter through this dialogue so Power BI adds the required metadata. Then select its generated table and use this complete replacement definition to give the choices plain business labels:

Metric = {
    ("Sales", NAMEOF('Sales'[Sales Value]), 0),
    ("Gross profit", NAMEOF('Sales'[Gross Profit]), 1),
    ("Margin", NAMEOF('Sales'[Gross Margin]), 2)
}

Use explicit measures, not automatic column aggregations. The trailing numbers define display order; they are not financial thresholds. Leave the generated parameter table disconnected. Microsoft: field parameters.

Connect one chart and make the choice obvious

Insert a clustered bar chart. Put Sales[Channel] on the Y-axis and the Metric parameter on the X-axis, replacing any fixed measure already there. Configure the parameter slicer for single selection, turn off Select all and select Sales as the saved default.

Title the slicer “Choose a measure”. Enable data labels, use automatic axis bounds and set display units to None for this small test. Avoid a fixed currency axis title: use “Selected measure” and show the active measure prominently beside the chart. Test each currency and percentage format in the actual visual.

For a dynamic title, create the following measure. It reads the generated order column rather than the parameter display column. Set the chart title’s conditional formatting to Field value and select Chart Title.

Chart Title =
VAR MetricOrder = SELECTEDVALUE('Metric'[Metric Order], -1)
RETURN
    SWITCH(
        MetricOrder,
        0, "Sales by channel (£)",
        1, "Gross profit by channel (£)",
        2, "Gross margin by channel (%)",
        "Select one measure"
    )

This title mapping follows the parameter’s three ordering values. If you change those entries, update the mapping in the same release. SELECTEDVALUE supplies the fallback when there is not one distinct order value. Microsoft: SELECTEDVALUE.

Prove every selection before rollout

Build a temporary table with Channel and all three measures. Compare it with the chart after each selection. Test channel filters, no-data selections and saved report defaults. Decide whether bars stay in channel order or reorder by the selected measure; explain that choice to readers.

Use a separate test copy to add a zero-sales row and confirm that margin is blank. Try clearing the parameter filter too: Power BI treats no field selection as all fields, not as “show nothing”. Check bookmarks and reset behaviour so they cannot unexpectedly restore mixed units.