Who's driving growth, and can I trust the join?
Golden reads every sheet in your workbook, infers how they relate, and stops you before a join quietly inflates a number.
Five sheets and a VLOOKUP habit
Theo runs RevOps at Northstar Office Supply. Sales exports the order system as a workbook with five sheets: orders, order items, customers, products and regions.
He needs growth by segment, margin by product line and shipping SLAs by region. Flattening it by hand takes an afternoon, and the flat file gets the shipping average wrong anyway.
Relationships, inferred and guarded
Golden finds the relationships for you, from foreign keys when the source has them and from AI inference over column names when it doesn't. You check them in a diagram and fix any by hand.
Then every chart joins only what it needs, and a fan-out guard refuses any breakdown that would count the same order twice.
How Golden answers it
Load the workbook, check the model
Pick all five sheets. The Relationships view shows four joins tagged AI Inferred, each many-to-one, with orders and order items marked as fact tables.
Chart revenue by segment
Revenue lives on order items, segment on customers. Drop both on a stacked area and Golden joins two hops toward the one side, so nothing duplicates.
Plot margin against revenue
A scatter of margin % against revenue by sub-category. Accessories sits up top; Laptops sits far right and near the floor.
Hit the fan-out guard
Average ship days by region is safe. Drag sub-category onto color and Golden refuses, explaining that each order would repeat once per line item and inflate the average.
Growth, margin and an honest average
- 01H1 revenue grew 20.1%, from $2.13M to $2.56M.
- 02Mid-Market added $449,711, which is 105% of the growth. SMB slipped $22K and Enterprise was flat to the dollar.
- 03Accessories earns a 54.5% margin on $1.83M; Laptops bring $1.52M at 15.2%.
- 04Orders ship in 4.62 days on average. The flattened file says 5.33, because bigger orders ship slower and get counted once per line.
Take it to the team
Run Explain Change on revenue, H1 2024 against H1 2025, to break the growth down by segment and then sub-category. Forecast each segment to December to see where Mid-Market ends the year.
Share the dashboard with the Segment slicer preset to Mid-Market, and export the margin table to Google Sheets for merchandising.
Bring the whole workbook
Load every sheet, let Golden find the joins, and trust the totals.