Datasets / Generated

Subscription cohorts

Monthly signup cohorts with retention and MRR by plan.

Ask a question of it Download CSV Jupyter notebook

1,056 rows6 columns Generated

Worked examples

Each of these is a real run: the pipeline picked the columns, shaped the rows and chose the form. Open the trace to see what it decided and why.

Accounts still active, by months since signup

Every cohort is put on the same x axis - months since signup, not calendar months - which is the only way curves of different ages can be compared.

what does retention look like by months since signup? — 19 rows charted from 1,056, drawn as a area

Ask this yourself See the working

How it decided

Finding the data117 ms

Subscription cohorts: 1,056 rows, 6 columns. months_since_signup and active_accounts, summed over every cohort.

Shaping the data3 ms

1,056 rows in, 19 out across 2 columns. Every cohort is put on the same x axis - months since signup, not calendar months - which is the only way curves of different ages can be compared.

Choosing the chart2 ms

area. A single decaying series reads as an area: the fill carries the magnitude that is being lost, which a bare line does not.

Applying defaults1 ms

Palette, spacing, axis titles and legend placement applied. 0 issues found.

Share of monthly recurring revenue by plan

A share of a whole with three parts is one of the few honest uses of a pie; four or more slices and this becomes bars.

how does MRR split across plans? — 4 rows charted from 1,056, drawn as a pie

Ask this yourself See the working

How it decided

Finding the data114 ms

Subscription cohorts: 1,056 rows, 6 columns. plan and mrr for the most recent activity month only.

Shaping the data7 ms

1,056 rows in, 4 out across 2 columns. A share of a whole with three parts is one of the few honest uses of a pie; four or more slices and this becomes bars.

Choosing the chart11 ms

pie. Parts of one total, few enough to read as slices. Every slice is labelled, because angle alone is not a readable quantity.

Applying defaults1 ms

Palette, spacing, axis titles and legend placement applied. 0 issues found.

Schema

6 columns, typed as the pipeline sees them — which is how it knows what can go on a time axis and what can be summed.

Column	Type	Example
cohort_month	date	2024-01-01
activity_month	date	2024-01-01
months_since_signup	number	0
plan	text	Free
active_accounts	number	115
mrr	number	0.0

First 8 rows

cohort_month	activity_month	months_since_signup	plan	active_accounts	mrr
2024-01-01	2024-01-01	0	Free	115	0.0
2024-01-01	2024-01-01	0	Starter	67	1273.0
2024-01-01	2024-01-01	0	Pro	47	2303.0
2024-01-01	2024-01-01	0	Team	20	2980.0
2024-01-01	2024-02-01	1	Free	95	0.0
2024-01-01	2024-02-01	1	Starter	55	1045.0
2024-01-01	2024-02-01	1	Pro	39	1911.0
2024-01-01	2024-02-01	1	Team	16	2384.0

Ask it something else

what does retention look like by months since signup?how does MRR split across plans?which cohort retained best?

Start free