For data people will look at.
Tests don't care whether data looks real. Dashboards, demos and classes do.
I make…
Building an app instead?
Analysts · BI builders · portfolio builders
A dashboard with something to find.
- Findings you choose
- Up to 1M rows for Power BI
- Not the dataset every portfolio uses
Same columns. Same 20% churn. Only one has a finding.
Nothing to present.
Monthly plans churn 3× more.
You type“5,000 subscribers. Monthly plans churn 3× more. Spend rises with seats.”
Preview this dataset →Pre-sales · consultants · product marketers
Demos that land the point.
- The insight is guaranteed to be there
- No customer data involved
- Linked tables, so the app works end to end
You type“10,000 insurance claims. 3% are suspicious and cluster at one repair shop.”
Preview this dataset →Lecturers · bootcamps · trainers
A different dataset for every student.
- Plant the finding they must discover
- A fresh copy each, so no answer-sharing
- Mess to clean, at the level you set
You type once“300 students. Attendance predicts the final grade. 5% of marks missing, a few duplicate rows.”
| id | attend | grade |
|---|---|---|
| S-0417 | 60% | 51 |
| S-0418 | 83% | 60 |
| S-0419 | 94% | missing |
| S-0420 | 60% | 32 |
r = 0.81
| id | attend | grade |
|---|---|---|
| S-1102 | 62% | 37 |
| S-1103 | 46% | 38 |
| S-1103 duplicate | 46% | 38 |
| S-1104 | 97% | 69 |
r = 0.90
| id | attend | grade |
|---|---|---|
| S-2250 | 64% | 56 |
| S-2251 | 63% | 46 |
| S-2252 | 81% | 64 |
| S-2253 | 78% | missing |
r = 0.84
- Same planted finding in every copy
- Mess to clean, different in each