For the complete documentation index, see llms.txt. This page is also available as Markdown.

Repurchase / churn / LTV

Cohort analysis groups people who made their first purchase at the same time and tracks how the customers in that group keep behaving afterwards. Changes along the horizontal axis (rows) show how customers who first purchased at a given time change their behavior in the following periods. Changes along the vertical axis (columns) show whether values go up or down regardless of first purchase time, so you can check whether the service is improving overall.

What is a cohort chart?

  • A cohort means a ‘group of the same kind’. Cohort analysis means grouping customers into groups that share a specific criterion and analyzing them.

  • The important question is how to set that criterion. There are many options, but by default Laplace defines it as ‘people who made their first purchase in your store at the same time’ (First-purchase cohort).

  • One of the most important parts of running a commerce business is how customers' purchase journeys unfold and how to design them. That is why the ‘first purchase’, the start of the purchase journey, matters most, and grouping customers who first purchased at the same time into one group (cohort) and following their trend is very meaningful.

You said people who made their first purchase are grouped together. Does the number of people in that group never change?

Correct. If 100 customers made their first purchase in May 2022, their trend is followed indefinitely from that month as M+1 (June), M+2 (July), and so on. Whatever M+ month it is, the denominator for the repurchase rate stays at 100 people. The Cohort users column of the cohort table shows this number.

How repurchase rate and churn rate are calculated

Cohort repurchase rate and churn rate are calculated based on groups of customers whose first purchase happened at the same time.

  • Cohort users: The number of users who made their first purchase in the same period. It is used as the denominator for every elapsed period and does not change afterwards.

  • Repeat buyers: The number of cohort users who purchased again in the chosen elapsed period and cumulative setting.

  • Churned users: The number of cohort users who did not repurchase under that criterion.

Repurchase rate and churn rate formulas

  • Repurchase rate = Repeat buyers / Cohort users × 100

  • Churned users = Cohort users - Repeat buyers

  • Churn rate = Churned users / Cohort users × 100 = 100 - Repurchase rate

The period in which repurchases are counted depends on the unit (daily, weekly, monthly and so on) and the chosen elapsed period. Also, the repeat buyers included in each interval depend on which cumulative setting you choose: Not cumulative, Cumulative (on or before), or Cumulative (on or after). Do not directly compare repurchase and churn rates from charts with different settings.

Reading the repurchase rate

The repurchase rate cohort chart tracks the repurchase rate each cohort generates over time. Cohorts (groups of the same kind) are split by first purchase time. Users are grouped by month of first purchase, and the repurchase rate of each cohort is tracked. In the cohort table below, each row is a first purchase period (cohort), the Cohort users column is the number of users who made their first purchase in that period, and the M + 1, M + 2 … columns are the repurchase rates for each period elapsed since the first purchase. The Average row at the top is the weighted average of all cohorts.

Repurchase rate formula

The repurchase rate is calculated as Repeat buyers / Cohort users (first-purchase users) × 100.

Cohort table: Cohort users column and M + 1 column

For example, if 585 users belong to the June 2021 first-purchase cohort and 71 of the 585 repurchased in M+0 (June 2021), the repurchase rate is 12.14%. If 82 users repurchased in M+1 (July 2021), the repurchase rate is 14.02%. In both cases the denominator is the same 585 people.

In the table above, 3,400 users made their first purchase in November 2024, and 33.52% of them purchased again one month later (M + 1). The darker the cell, the higher the value.

[Caution]

  • Be careful when interpreting the last period. If you are analyzing the cohort chart on October 15, 2021, October 2021 has not finished calculating its cohort metrics, so it should be excluded from interpretation.

  • The same applies to every cohort area pointing at October 2021. The M+1 metric for users who made their first purchase in September 2021 is incomplete until October 31 has passed. So you may need to exclude the cohort areas on the diagonal from the last cohort area from interpretation. The table shows these cells in gray, and periods that have not arrived yet as -. [See the image below]

    Cohort table: gray cells (periods not finished yet) and the note below

[Choosing the unit]

Laplace offers five units for cohort analysis: daily / weekly / monthly / quarterly / yearly.

To choose the best unit, first calculate the purchase cycle in days. Then find the unit with the smallest distance in the table below.

The closest unit is found with distance = |purchase cycle - days per unit| (absolute value).

If the purchase cycle is 49 days, the distances are as follows.

Daily
Weekly
Monthly
Quarterly
Yearly

Days per unit

1 day

7 days

30 days

90 days

365 days

Distance

48 days

42 days

19 days

41 days

316 days

Based on this result, you should analyze monthly cohort metrics. If distances are equal, it is better to follow the smaller unit.

(See also: How to calculate the purchase cycle )

Reading customer lifetime value

The customer lifetime value cohort chart tracks the revenue/margin one user in each cohort generates over time.

Cohorts (groups of the same kind) are split by first purchase time. Users are grouped by first purchase time, and the lifetime value of each cohort is tracked.

Customer lifetime value formula

Customer lifetime value is calculated as cumulative revenue / first-purchase users per cohort.

On screen, when Data is Order data, you see it as Lifetime revenue (LTR) in Metric (cumulative revenue / cohort users). With Base product order data you see it as LTV (cumulative contribution margin / cohort users).

For example, 585 users made their first purchase in June 2021, and revenue and cumulative revenue these 585 users generated in M+0 (June 2021) was KRW 27,896,250, so the lifetime value is KRW 47,686.

In M+1 (July 2021), 42 users generated revenue of KRW 4,981,378, for cumulative revenue of KRW 32,877,628. So the M+1 lifetime value is KRW 56,201.

M+1 cumulative revenue = M+0 revenue + M+1 revenue = KRW 27,896,250 + KRW 4,981,378 = KRW 32,877,628 M+1 lifetime value = KRW 32,877,628 / 585 users = KRW 56,201

Lifetime value is cumulative, so does it always increase toward the right of the graph?

Yes. Lifetime value is the average purchase amount per customer, so it can only stay the same or grow toward the right. If you want to see how much it grew each month, choose LTR diff in Metric. The difference shows how much it increased compared with the previous point.

[Caution]

  • Be careful when interpreting the last period. If you are analyzing the cohort chart on October 15, 2021, October 2021 has not finished calculating its cohort metrics, so it should be excluded from interpretation.

  • The same applies to every cohort area pointing at October 2021. The M+1 metric for users who made their first purchase in September 2021 is incomplete until October 31 has passed. So you may need to exclude the cohort areas on the diagonal from the last cohort area from interpretation. These are the gray cells in the table.

[Choosing the unit]

Laplace offers five units for cohort analysis: daily / weekly / monthly / quarterly / yearly.

To choose the best unit, first calculate the purchase cycle in days. Then find the unit with the smallest distance in the table below.

The closest unit is found with distance = |purchase cycle - days per unit| (absolute value).

If the purchase cycle is 49 days, the distances are as follows.

Daily
Weekly
Monthly
Quarterly
Yearly

Days per unit

1 day

7 days

30 days

90 days

365 days

Distance

48 days

42 days

19 days

41 days

316 days

Based on this result, you should analyze monthly cohort metrics. If distances are equal, it is better to follow the smaller unit.

(See also: How to calculate the purchase cycle )

How to use the chart

For common board features such as the date range, filters, and card menu, see Using the dashboard › Analysis board guide.

  1. In the left rail, click Dashboard › in the navigation panel, Cohort Analysis › Repurchase Rate / LTV.

    Repurchase Rate / LTV board
  2. In the first row of the filter bar, choose the period, unit, and refunds/cancellations.

    Filter bar: period, unit, refunds/cancellations
  • Period: By default you see the last 24 months. The number of rows (cohorts) in the table follows this period.

  • Unit: Choose from Daily, Weekly, Monthly, Quarterly, and Yearly (Monthly by default). When you change the unit, the period also changes to that unit's default (daily 7 days, weekly 25 weeks, monthly 24 months, quarterly 5 quarters, yearly 2 years).

  • Refunds and cancellations: Excl. refunds/cancellations by default. Switch to Incl. refunds/cancellations to also count refunded and canceled orders.

  1. In the What do you want to see? row, choose Data, Metric, and Break down by.

    What do you want to see? row: Metric list
  • Data: Choose Order data (default) or Base product order data. 👉 Base product order data needs additional setup. See the Base product data guide for details.

  • Metric: Repurchase Rate by default. With Order data you can choose Repurchase Rate, Repurchase payment rate, Churn Rate, LTR, LTR diff, Actual LTR, Actual LTR diff, Repurchase User count, Repurchase payments, Churned users, Revenue, Actual Sales, Repurchase sales, Repurchase actual sales, Repurchase ARPPU, Repurchase actual ARPPU, Acc Revenue, and Cumulative actual sales. Base product order data adds metrics such as LTV, Contribution Margin, Gross Margin, COGS, and Ads Spend.

  • Break down by: Split cohorts by All (no split) (default), Shop, Sales channel, Brand, or Product. With Base product order data, split by Sales channel, Base product brand, or Base product name.

  1. In the Only show what? row, use Store and Brand to keep only the orders you want to see. Add other conditions with Advanced filter.

    Only show what? row: Store, Brand, Advanced filter
  2. In the Cohort trend card header, choose a view: Average, Trend, or Cards.

    Cohort trend card: Trend and elapsed period
  • Average: Draws the weighted average of all cohorts for each elapsed period (M + 0, M + 1 …).

  • Trend: Connects the value of the chosen elapsed period (for example M + 1) across cohorts (first purchase periods). Use it to check whether recent cohorts are getting better.

  • Cards: Shows M + 1, M + 3, and M + 6 values as cards and compares the latest cohort with the average.

  1. Click the analysis conditions in the card header (First-purchase cohort, Calendar, Not cumulative) to change how values are calculated.

    Analysis conditions: retention method list
  • Cohort basis: Choose First-purchase cohort (default), 2nd purchase cohort to 5th purchase cohort, or Purchase cohort (grouped into the period of every purchase).

  • Retention method:

    • Calendar (by calendar month): Calculates elapsed periods by the calendar. ex) If a customer purchases on May 20 and repurchases on June 1, it counts as M+1 retention.

    • Rolling (by days elapsed): Calculates elapsed periods by the time it took for retention to happen. ex) If a customer purchases on May 20 and repurchases on June 1, retention took 12 days, which is within 30 days, so it counts as M+0 retention.

    • Frequency (by purchase count): Calculates the conversion rate for each purchase count (2nd purchase = count 1, 3rd purchase = count 2 …).

  • Cumulative setting: Not cumulative counts only customers who purchased again in that period, Cumulative (on or before) counts customers who purchased again at least once up to that period, and Cumulative (on or after) counts customers who purchased again at least once from that period to now. Shown only for repurchase and churn metrics.

  1. When you split cohorts with Break down by, two more conditions appear in the card header.

    Analysis conditions: N of A, A → A

Nth purchase cohort condition

  • N of A: Sets the customer's Nth event among events where analysis unit A occurred as the cohort (ex. measured from the first event of each product A, B, C)

  • A of N: Sets the customer's Nth event as the cohort regardless of analysis unit (ex. product A bought by a first-purchase customer)

Retention condition

  • A → A: Customers who bought product A and then bought A again

  • A → Any: Customers who bought product A and then bought any product, including A

  • A → NOT A: Customers who bought product A and then bought a product other than A

Churn rate / LTV screen guide

The Churn rate / LTV screen shows expected churn rate and expected LTV for each unit period, based on the churn window (N cycles) rather than cohorts. For common board features such as the date range, filters, and card menu, see Using the dashboard › Analysis board guide.

  1. In the left rail, click Dashboard › in the navigation panel, User Analysis › Churn rate / LTV.

    Churn rate / LTV board
  2. In the first row of the filter bar, choose the period, unit, and Churn window.

    Filter bar: period, unit, Churn window
  • Unit: Daily, Weekly, Monthly, Quarterly, Yearly (Monthly by default)

  • Period: By default you see the last 3 months (the 1st of two months ago to today).

  • Churn window: Users who purchased N cycles ago and have not purchased again are counted as churned. The default is the last 3 cycles (3 months when the unit is Monthly). Adjust it to your repurchase cycle.

  • In the Only show what? row, you can narrow down by Store and Brand.

  1. In What do you want to see?, choose the metrics in Metric and the Chart type. The table under the chart shows the same values by period.

    Churn rate and LTV card: chart and table below
  • By default only Expected churn rate is shown. You can add Expected LTV, Repurchase Rate, Paying users (N cycles), Churned users, Repeat buyers, and ARPPU (N cycles).

  • Expected LTV = ARPPU (N cycles) ÷ Expected churn rate. The lower the churn rate, the higher the LTV.

  • Expected churn rate and Repurchase Rate are shown with %.

  • Chart can be Line, Bar, Combo (bar + line), or Card view.

Ask your dedicated manager

Having trouble using the dashboard?

Contact us through the inquiry page below. Your manager will help you.

💬 Contact us on our website

Email: product@laplacetec.com

Last updated