Sales Performance Analytics Dashboard (Power BI)

Problem

The brief was to turn raw retail sales, returns and customer files into a report an executive could use to answer everyday questions: how revenue, orders, profit and returns are trending, how they compare with a target, which categories, products, regions and customer segments drive them, and what would happen if prices changed. The source was a public sample retail dataset (roughly 25,000 orders, 17,000 customers and 290 products over three years) used during Power BI certification study, so this is a practice project: it demonstrates modelling and report-design skills rather than production work for an employer.

Constraints

Sales arrived as several files that had to be combined, alongside separate lookup files for products, customers, territories and dates, so the data needed a clean, consistent model before any visual could be trusted. Every measure had to stay correct under any combination of slicers and filters, and the same definitions of revenue, profit and return rate had to be used on every page. The report also had to work at two levels: a one-page executive summary, and detail pages for products and customers that people could reach with consistent navigation.

Approach

I loaded the source files with Power Query, combining the multiple sales files from a folder into a single sales fact table and loading returns as a second fact table beside it. The lookup files became dimension tables joined to both facts with single-direction many-to-one relationships; a second date relationship (stock date) is kept inactive so that order date drives time analysis by default. All measures live in a dedicated measure table, built in layers: base measures (orders, customers, quantity sold and returned, revenue, cost, profit), ratios (return rate, share of all orders and returns, average revenue per customer), time intelligence (year to date, previous-month comparisons, rolling 10-day revenue and 90-day profit), targets with gap measures, and a what-if price adjustment that recalculates revenue and profit. Detail measures for customers only return values when a single customer is selected. On top of the model I built the report pages: an executive dashboard, a map, product and customer detail pages, a report-page tooltip, and analytical pages using Q&A, a decomposition tree and key influencers.

Outcome

A multi-page interactive report where KPI cards, trend lines, gauges, maps, tables and analytical visuals all read from the same shared measures, so they agree with each other under any slicer selection. Targets are calculated by rule (a fixed uplift on the previous month), so gap-to-target indicators and arrows update automatically as the data changes. Action-button navigation and bookmarks let a viewer move from the executive summary to the product and customer detail pages without leaving the report.

What I'd do differently

I would add reconciliation checks between each target measure and the base measure it depends on, and reconcile every KPI back to the source totals before sharing the report. I would also load the model from a governed, scheduled source instead of local files, and add row-level security if the report served more than one audience.

Portfolio note: this project uses a public sample retail dataset and was built while studying for the Microsoft Power BI Desktop certification. It shows my modelling, DAX and report-design approach, and is not client or employer work.

Data model

The two fact tables share the same calendar, product and territory dimensions, which is what allows a measure like return rate (returns divided by sales) to be sliced by product, region or month consistently. All ~40 DAX measures sit in a separate measure table, built from base measures upwards, so each definition is written once and reused everywhere.

Report structure

• Executive dashboard: revenue, orders and returns as KPI cards, a revenue trend, monthly breakdowns and orders by category, with slicers and page navigation.
• Map: performance by country and continent.
• Product details: revenue, orders and profit gauges against targets, plus profit and returns trends.
• Customer details: orders by income level and occupation, and a top customers table that shows per-customer detail on selection.
• Category tooltip: a report-page tooltip showing weekly orders when hovering a category.
• Analytical pages: Q&A, a decomposition tree and key influencers for exploring the drivers behind the headline metrics.