Carmen's review and iteration

Step 1: Dashboard purpose statement

Before showing Carmen the dashboard, write a purpose statement. Who is this dashboard for? What questions does it answer? What does it not cover?

Open your analysis template and fill in the Dashboard Purpose Statement section:

• Audience: Carmen (weekly performance review) and her sales manager (channel-level planning)
• Questions answered: Weekly revenue by channel and beer style, trends over 12 months, filter-driven drill-down
• Questions not answered: Production volumes, inventory levels, individual bar/restaurant performance, profitability (costs are in the data but not yet in the dashboard)

A dashboard without a stated purpose gets used by people it was never designed for. Carmen's investors might want quarterly summaries with year-over-year growth -- different audience, different dashboard. The purpose statement draws the boundary.

Step 2: Present to Carmen

Send Carmen the dashboard link and a summary of the key findings. Keep it concise -- Carmen reads fast and prefers quick hits:

• Total revenue across all three channels (the unified number she's been asking for)
• Which channel contributes the most revenue
• Top three beer styles by revenue
• The trend: is revenue growing, flat, or declining month over month?
• The seasonal patterns: which styles are only available in certain quarters

Carmen responds within minutes:

omg this is beautiful. I've been asking for this for TWO YEARS

Then:

ok one thing -- can we also add a comparison to the same period last year? My investors keep asking about year-over-year growth and I have to scramble every time

Step 3: Handle the year-over-year request

Carmen's YoY request is reasonable and fits within the dashboard paradigm. You have twelve months of data from 2024, so you can show month-over-month trends within the year.

For a true year-over-year comparison, you'd need 2023 data too. Since you only have 2024, you can:

• Add a panel showing monthly growth rates (each month vs previous month)
• If the data has enough range, show quarter-over-quarter comparison
• Note to Carmen that full YoY requires 2023 data she hasn't provided yet

Direct AI to add a growth panel to the dashboard:

Add a panel to the Metabase dashboard showing month-over-month revenue growth rate as a line chart. Use the same revenue definition as all other panels. Place it below the trend line.

The new panel must use the same revenue definition. This is the governance test -- Carmen's scope addition inherits the governed definition. If AI generates the growth panel with a different revenue formula than the other panels, you've lost consistency.

Handle the seasonal styles: if Summer Wheat has zero revenue in Q4, the month-over-month growth rate is undefined, not -100%. Direct AI to handle these cases.

Step 4: Scope management

Carmen also mentions:

oh and eventually it would be amazing to see our actual brewing volumes next to the sales -- like, are we brewing too much of something that doesn't sell?

This is a different project. Production data is a separate source with different structure, different questions, and different definitions. The dashboard you've built answers sales questions. Adding production would change what the dashboard is about.

Respond professionally: "That's a great idea and it would be a separate project -- the production data brings a new set of questions about capacity planning and waste. I'd recommend keeping this dashboard focused on sales performance and building a production dashboard separately."

Carmen will accept this easily. She knows the production data is a different world.

Step 5: Final verification and push

Verify the dashboard with the new growth panel:

• Do all filters still work with the added panel?
• Does the growth panel use the same revenue definition as the KPI and breakdown panels?
• Are seasonal styles handled correctly in the growth calculations?

Once verified, commit and push to GitHub:

Commit all work with a descriptive message summarising the deliverable: unified sales dashboard for Bona Lluna Cervesa with three-channel reconciliation, governed metric definitions, and interactive filters.