How Yuktikara runs on data

“What were our sales?” sounds simple. The data holds at least four believable answers, and only one of them is right. The gap between the first and the last is 20.3%. The definition belongs in one place, a measure in the semantic model, and the ontology holds the pieces it works on: order status, line amounts, and accepted returns with the day they were accepted.

1. Gross, every order$13.88M

   Counts cancelled and pending orders, and ignores markdowns and returns

2. Order total$13.02M

   Only real sales, but includes sales tax

3. Subtotal$12.03M

   Right orders, no tax, but forgets the returns

4. ✓ Net sales$11.06M

   Completed and Shipped orders, minus accepted returns on the day they're accepted. No tax

Sales is only the most familiar example. The same data covers buying, stock month by month, promotions and each store’s plan, so it also answers which supplier is letting us down, which shelves need refilling, whether the clearance paid, and which stores are behind plan. A platform built for one question answers one question.

Published snapshot, Jan 2025 to Aug 2026. The refresh notebook regenerates the data up to yesterday, so its figures move, but the gap stays close to 20%.