Weekly cohort analysis

Q: What is cohort analysis?

Cohort analysis groups users by when they joined and tracks how each group behaves over time. It reveals whether the product is genuinely getting better — not just whether top-line numbers are growing because of new signups.

Q: How often should you run cohort analysis?

Weekly for product teams, monthly for executives. Weekly cadence catches the cohort movement that comes from specific product changes; monthly is enough for strategic reviews.

Weekly AI cohort analysis. Tracks each signup cohort's retention curve over time and highlights when product changes shift the curve.

Use this template How automations work

Runs weekly Scheduled Delivered to Slack, Email

Weekly cohort analysis

Active

Trigger

Runs weekly

Instructions

Generate a weekly cohort retention report.

Delivery

Slack · Email

Example output

A realistic sample of what this automation delivers — your version draws on your data.

Weekly cohort analysis Runs weekly · Sent to Slack, Email

Sample output

Cohort retention, last 4 cohorts % retained

Cohort 12wk ago Cohort 8wk ago Cohort 4wk ago This week

Cohort retention — last 12 weeks

Cohort	W1	W2	W4	W8	W12
Week 12 ago	100%	71%	54%	42%	41%
Week 8 ago	100%	74%	56%	46%	—
Week 4 ago	100%	78%	61%	—	—
This week	100%	—	—	—	—

Improving. W4 retention is up 7 pts in 8 weeks — a clear trend that started the week of the onboarding redesign.

Below median. The week-of-Apr-21 cohort underperformed by 9 pts — that was the week of the failed deploy. Marketing-sourced cohorts on that week didn’t get the new onboarding.

Tie. This week’s W1 came in at 78%, matching the all-time best. Next week’s data point matters.

The prompt

Copy, paste, customize.

Drop this into a Basedash automation. AI fills in the numbers from every source you've connected.

Instructions

Generate a weekly cohort retention report.

For each weekly signup cohort over the past 12 weeks, compute retention at week 1, 4, 8, and 12.

Surface:

**Cohort table.** A 12-row × 4-column retention table.

**Standouts.** Cohorts that beat the trailing 12-week median by > 10% — what product or marketing changes happened that week?

**Concerns.** Cohorts that underperformed by > 10% — same question.

**Channel cohort.** Same analysis cut by acquisition channel.

Close with one product change worth running an A/B test on, based on cohort movement.

Data sources