A classic transactional sales table: one row per sale with a date, region, product, channel, units and revenue. The shape that every BI tutorial, pivot-table demo and dashboard example assumes.
The small files are static downloads. The large ones are generated in your browser from the same fixed seed, so every copy of sales-100000.csv on earth is byte-identical — reproducible test data with no 60 MB download.
→ Open this dataset in the dashboard builder · → Open in the CSV editor
| date | region | product | channel | units | unit_price | revenue |
|---|---|---|---|---|---|---|
| 2024-01-01 | North | Drift Clock | Retail | 3 | 41.59 | 124.77 |
| 2024-04-01 | South | Drift Clock | Retail | 9 | 172.39 | 1551.51 |
| 2024-07-01 | East | Drift Clock | Retail | 11 | 90.77 | 998.47 |
| 2024-09-30 | South | Cedar Shelf | Online | 4 | 180.60 | 722.40 |
| 2024-12-31 | Central | Meadow Chair | Retail | 5 | 63.42 | 317.10 |
| 2025-04-01 | North | Aurora Lamp | Retail | 5 | 43.12 | 215.60 |
| 2025-07-01 | Central | Tide Mug | Online | 9 | 70.43 | 633.87 |
| 2025-09-30 | West | Pebble Speaker | Online | 12 | 139.11 | 1669.32 |
| column | description |
|---|---|
date | Transaction date (ISO 8601), spanning two years |
region | Sales region |
product | Product name |
channel | Sales channel |
units | Units sold |
unit_price | Price per unit (USD) |
revenue | units × unit_price |
Two years of daily transactions across five regions, eight products and three channels. Revenue is exactly units × unit price, so aggregates check out — handy when you're verifying a pivot or a GROUP BY.
The mix is deliberately uneven (some products and regions sell more), so charts built on it look like real business data instead of uniform noise.
Good for: Pivot table and GROUP BY demos · Dashboard and chart examples · Testing imports that expect a date + amount shape.
License: CC0 / public domain — use it anywhere, no attribution needed.
Yes, exactly, on every row — so SUM checks and pivot validations come out consistent.
Yes. Dates spread monotonically across two years (2024–2025), so time-series charts work out of the box.
CC0 (public domain). Use it in tutorials, tests, courses, screenshots and products — no attribution required.
Yes — every size is generated from a fixed seed, so the same file is byte-identical for everyone, forever. Reproducible tests, stable teaching materials.