Reading cohort value without assuming future retention

Read cohort value as an observed amount through a stated customer age and adjustment cut-off. Divide by the original customer count and compare groups at the same age. Spending already recorded does not establish how many customers will return or what they will spend later.

Define the observed amount

Start with every qualifying entrant, including customers who never bought again. Sum the chosen value of that cohort's qualifying orders placed from entry through a fixed elapsed window. Divide by the original entrant count, not by the number who returned.

Observed value per entrant by day 90 = qualifying value of orders placed within 90 elapsed days of entry, including the entry order, with adjustments recorded by the stated cut-off ÷ original cohort entrants.

Day 90 is illustrative. Specify whether the measure uses product sales after discounts and relevant reversals, an order total including delivery and tax, or another amount. Keep currency and order-status rules consistent.

In Shopify, net sales deducts discounts and sales reversals from gross sales; total sales includes further components such as taxes and shipping. Those report labels do not automatically produce the custom order-cohort amount above.

A refund or cancellation may be recorded after the order window. State the later adjustment cut-off and mark the value provisional while changes can still arrive. Avoid deducting an adjustment again if the starting field already includes it. Sales per entrant is not contribution margin; a cost-based decision needs cost records and their coverage.

Steps to Calculate Observed Cohort Value

Compare equal histories

Suppose two hypothetical cohorts each have 100 entrants. If their qualifying, adjusted order values through day 90 are A$4,000 and A$5,000, their observed values per entrant at that age are A$40 and A$50. The figures say nothing by themselves about day-180 value, future repeat purchasing or margin.

Show cohort size, qualifying order count, entry dates, elapsed window, currency, value formula, extraction date and adjustment cut-off. Include only entrants whose full window has elapsed in a completed day-90 comparison. Keep incomplete periods blank or label them partial rather than filling them with assumed spending.

Separate recorded value from projections

A period value shows the amount associated with one interval. A cumulative value shows the amount through that interval. Name the view before comparing curves.

If a view includes projected values, keep projected cells outside any figure labelled observed value.

Check whether a few large orders dominate an average and whether cohorts differed in first product or offer. Treat those differences as context for investigation, not explanations established by the value grid. Refresh the observed result after further orders and adjustments arrive.

Key Considerations for Valid Cohort Analysis