MKT 626 | Class 5 | Arbor Threads
Cohort Unit EconomicsWhen does a cohort pay back its costs?
Take one real Arbor Threads cohort and build its discounted cash flow the way the class workbook does: start from revenue per active customer, scale by % active to get revenue per cohort member, scale by margin to get cash flow, discount it, then subtract what you paid to acquire them. Time runs in quarters since acquisition, and quarter 0 is the acquisition quarter itself.
Assumptions and cohort
trlog_cds_5yr.csv | time = quarters since acquisition | quarter 0 = the acquisition quarterOne cohort, one row of the workbook at a time
Cumulative expected DCF per member
Each quarter stacks its discounted expected cash flow on the pile.
The chart is this table
Every bar above is one row below, exactly the workbook's columns.
Reading the race
Every line is one cohort's cumulative net expected DCF per member, aligned by quarters since acquisition and using the sliders above. A line ends where the data ends: the 2023-Q4 cohort has one observed quarter, not worse economics. Crossing $0 is payback. Hover a line or its row to pick a cohort out.
Every cohort under the same assumptions
PV and net are per member over the observed window only.
Model, conventions, and source notes
Workbook logic. This is the sheet Class 5 - DCF in Cohort Unit Economics made live. Time runs in quarters since acquisition; quarter 0 is the acquisition quarter itself. The ladder, per quarter: revenue per ACTIVE customer; times % active = expected revenue per newly acquired customer (equivalently, cohort revenue / cohort size, so revenue per cohort MEMBER); times contribution margin = expected cash flow per member; times the discount factor = expected discounted cash flow, at the quarterly rate implied by the annual slider, (1 + annual)^(1/4) - 1. Quarter 0 is not discounted (factor 1.000), and CAC is paid at quarter 0. ROI is net gain per member over CAC, the workbook's framing.
Data. Aggregated from trlog_cds_5yr.csv, the Arbor Threads 20% customer sample you work with, cohorts 2019-Q1 through 2023-Q4. The class sheet's worked example uses the full customer file, so its 2022-Q1 cohort is about five times this size (4,576 members vs 908) with slightly different per-member revenue ($402 vs $382 in period 0). Sampling moves the counts, not the economics. Observed windows are censored: a cohort acquired later simply has fewer quarters of history.