Excel & Power BI for Finance: Finance Dashboard End-to-End: Star, DAX, Layout

Download the AtlasParts dataset and the Episode 12 Power BI starter file — follow along in your own copy of Power BI Desktop: github.com/GoCelesteAI/excel-powerbi-for-finance.

This is the finale. Eleven episodes collapse into one page. We take the AtlasParts star schema from Episode 9, the measures from Episodes 10 and 11, and assemble the dashboard a finance team’s CFO actually looks at on Monday morning.

The lesson here is layout, polish, and interaction. The DAX is already done. The model is already done. What’s left is the part most tutorials skip: making it readable, making it survive a slicer click, and making it feel like a product instead of a developer’s scratch pad.

What goes on a finance dashboard

The hardest part of any dashboard isn’t the visuals. It’s deciding what doesn’t go on it.

Every finance team gets asked the same five questions, in roughly this order. How are we tracking versus plan. How does this year compare to last. What’s the monthly trend. Which regions or products are dragging. Where’s the cash sitting.

A dashboard answers these five at a glance. Anything else belongs on a drill-through page or a different report entirely. That maps to a layout pattern almost every production finance board follows: a KPI row at the top, a trend chart below it, a breakdown by region or product on one side, AR aging or budget-versus-actual along the bottom. Four zones. Five questions. Everything else cut.

The KPI row

Across the top of the page sit four cards. Revenue YTD. Gross Margin Percent. OPEX. EBITDA. Each is one measure, already built in Episodes 10 and 11.

A naked card visual with just a number is a missed opportunity. What makes a KPI card readable is the delta — the comparison number underneath. Revenue YTD beside “+8.4% vs prior year” tells you twice as much as the bare number. The delta gets conditional formatting: green up-arrow for positive, red down-arrow for negative.

In the simulator we use the shared KPICard component with a delta prop. In real Power BI it’s the New Card visual (or the multi-row card) plus a delta measure plus conditional formatting on the value colour. Either way, the result is a four-card strip where the row of arrows is the first thing the reader processes.

A useful default: when one of the four KPIs would mean nothing without context, replace it with a more telling pair. Revenue YTD versus Plan, instead of Revenue YTD by itself. The point is the row reads — not what’s technically possible.

The revenue trend

Below the KPI strip, the headline chart. A line chart with twelve months on the x-axis and two lines: Revenue YTD this year, and Revenue YTD prior year. Solid line for current, dashed for prior. This is the chart that anchors every finance dashboard ever built.

The component is LineChartVisual with the multi-series API we added in Episode 11. Two LineSeries entries — one for 2026, one for 2025 — sharing a categories axis of month labels. The legend renders automatically when more than one series is present.

In real Power BI, this is a Line Chart visual with Calendar Month on X-axis and both Revenue YTD and Revenue YTD PY on the Y-axis values shelf. Power BI auto-colours the second series; manually setting it to a 60% grey dashed line makes the prior-year visually subordinate, which is what you want — the eye should track the current year and use prior year as background reference.

One chart, two measures, the most-used finance visual on Earth.

Margin by region

Next to the trend chart, gross margin by region. This one is a horizontal bar chart, not a vertical column chart. Region names like “North America” or “Asia Pacific” don’t fit under vertical columns without rotating the labels forty-five degrees, which makes them awful to read at a glance. Horizontal bars give each label its own line and the reader’s eye runs naturally from name to value.

The component is BarChartVisual with Gross Margin % on the value axis and Region on the category axis, sorted descending. In real Power BI it’s a Stacked Bar Chart visual with the same configuration.

Bar versus column matters more than people give it credit for. The rule of thumb: if your category labels are longer than three or four characters, use horizontal bars. If they’re calendar months or fiscal quarters with short labels, vertical columns are usually better because they match the time-running-left-to-right intuition.

AR aging

Aging buckets are a finance convention, not a chart tool default. Current. Zero to thirty days. Thirty-one to sixty. Sixty-one to ninety. Ninety-plus. A column chart with these five buckets reads as “where is the cash sitting” — and the longer the right-hand columns, the worse the cash position.

The interesting wrinkle here is where the bucket assignment lives. It’s not a measure, it’s a calculated column on the AR fact Table. Each AR row has a posted-date and an as-of-date; subtracting gives the age in days; a SWITCH expression on age slots the row into one of the five buckets.

AR Bucket =
SWITCH(
    TRUE(),
    [Days Outstanding] <= 0,  "Current",
    [Days Outstanding] <= 30, "1-30",
    [Days Outstanding] <= 60, "31-60",
    [Days Outstanding] <= 90, "61-90",
                              "90+"
)

This is the rare case where a calculated column is correct: the bucket assignment is intrinsic to the row, doesn’t depend on any user-applied filter, and you’ll filter and group by it constantly. Building it as a measure would force the bucket logic to evaluate per visual.

In real Power BI, the calculated column lives on the AR fact, and the visual is a Stacked Column Chart with AR Bucket on X and Total Outstanding on Y.

Budget versus actual

The bottom strip is a table. Quarter on rows, three columns — Budget, Actual, Variance Percent. With conditional formatting on the variance column, the table reads like a heat map.

Tables aren’t a fallback when you can’t think of a chart. They’re the right answer when the audience needs precise numbers and one number per row. CFOs reading a board pack want exact figures for budget conversations; a column chart of variances doesn’t help when the question is “what was Q3 actual revenue, exactly”.

The component is TableVisual with three columns and four to six rows depending on quarters in scope. Conditional formatting on the variance column: green for positive, red for negative, applied as a background colour rather than text colour so the heat-map effect reads from across the room.

Cross-filter — the signature Power BI moment

Click March on the revenue trend chart. Every other visual on the page responds. The KPI cards update to show just March’s numbers. The margin-by-region chart filters to March. The AR aging chart filters to March. The budget table highlights the row that contains March.

This is what cross-filtering means in Power BI, and it’s the single biggest reason a finance team picks Power BI over a spreadsheet. Excel can’t do this without VBA. Power BI does it for free, as long as the visuals share a model and the relationships are drawn correctly.

The simulator demonstrates this with CrossFilterStage — a wrapper that takes a script of timeline events and applies them to the visuals inside. The component layer handles the dimming, the source-highlight, and the filtered-by-other-visual badge. In a Remotion scene you just declare which visual is the source, which is filtered by it, and at what frame the click happens.

In real Power BI the behaviour is automatic. Most finance teams don’t realise they can click on a chart to filter the page. Add a small caption underneath the trend chart that says “click a month to filter the page” and discoverability improves dramatically.

The final dashboard

All the pieces assembled. KPI row across the top. Trend chart left, region chart right, AR aging below, budget-versus-actual table along the bottom. A region slicer in the top-right corner for users who prefer slicers to chart-clicks.

Eight visuals on one page, four zones, five questions answered. This is the artefact every previous episode of the season has been building toward.

The full Power BI Desktop chrome — title bar, ribbon, view rail, visualizations pane, fields pane, page tabs — wraps the canvas. The simulator uses PowerBIDesktop to render the entire window in one component; in real Power BI Desktop it’s the only window the user sees.

What we shipped

Four KPIs. Four charts. One slicer. Cross-filter wired across the whole page. One slicer. The CFO board pack a finance team can open every morning.

Plus the bigger arc: twelve episodes, from the first CSV extract to a refreshing Power BI dashboard. Six episodes on Excel — the bridge from CSV exports to spreadsheet workbooks finance teams actually use. Six episodes on Power BI — the model, the language, the dashboard. The pipeline from raw extract to refreshing CFO dashboard is now end-to-end and reproducible.

A few things to remember from the season. Power Query is the loading tier — eight Tables in, one star out. The star schema is the structural tier — fact at the centre, four dims around it, calendar marked as Date Table. DAX is the calculation tier — measures over filter context, time intelligence on top of the calendar. The dashboard is the presentation tier — four zones, five questions, cross-filter for free.

Each tier depends on the previous one being right. A bad model makes good DAX impossible. A bad calendar Table makes time intelligence silently wrong. A bad layout makes a dashboard nobody opens twice. Get each tier right, in order, and the next one becomes short.

That’s the whole season. Twelve episodes of building from CSV to dashboard. The finance team that owns this pipeline doesn’t have to ask the analytics team for anything ever again. That’s the bar.

What’s next

Excel and Power BI for Finance ships here. If you’ve followed all twelve episodes, you have a working AtlasParts dashboard, the full DAX library on GitHub, and the muscle memory to build the same pipeline against your own data. Future seasons will go deeper into specific finance domains — FP&A budgeting, cash flow forecasting, statutory reporting, M&A integration analytics. Subscribe to catch them when they ship.

Thanks for following the season. See you in the next one.