Sign in

Blog · Wallet share and penetration · Education

Adoption by purchase cohort: what institutions that bought three years ago hold now

How an education provider reads its base by first-purchase cohort, from the order history: the institutions that first bought in each year, how many are still customers, what programmes they hold now against what the cohort held at first purchase, the cohorts that expanded and the cohort that stalled, and why a cohort view finds the year something changed in onboarding or product that the renewal rate never explains.

The short answerGroup institutions by the year of their first purchase. For each cohort, count the institutions still buying, the programmes held now against the programmes held in year one, and revenue now against year one. A healthy cohort retains most of its institutions and holds more programmes each year. A cohort that retains fewer and holds the same one programme it started with stalled, and the year it started is where something changed. Cohorts are fixed at first purchase, and the sum across cohorts reconciles to the base.

An education provider's renewal rate is stable at 84 percent and has been for five years. Under it, the institutions that first bought four years ago hold three programmes each and the ones that first bought two years ago still hold one. Something changed between those years, and only the cohort view shows it. This guide sets out the cohort measures, the stalled cohort, and the identity.

The measures

Per cohort (first-purchase year):

Institutions at start; institutions still buying now; retention Programmes per institution at end of year one; now, among those still buying Revenue at year one; now Expansion ratio = programmes now ÷ programmes at year one

The rows you need

  • Orders: institution, programme, date, value.
  • Institution master: institution, entity, first purchase date.

Institution identifiers only.

The identity

Σ cohorts' institutions still buying = current customers Σ cohorts' revenue now = current revenue

Every institution in one cohort.

A worked cohort table

Cohort Started Still buying Retention Programmes, year one Programmes now Expansion Revenue now vs year one
2021 140 98 70% 1.3 3.1 2.4× 2.1×
2022 160 118 74% 1.2 2.6 2.2× 1.9×
2023 210 151 72% 1.1 1.3 1.2× 1.1×
2024 190 162 85% 1.2 1.4 1.2× 1.2×
2025 230 221 96% 1.1 1.2 too young

The 2023 cohort retained as well as its neighbours and did not expand. Its institutions hold what they bought and nothing more, two years on. Something in the 2023 onboarding, or the product that year, or the sales motion, left them as one-programme customers, and the 2024 cohort looks the same. The renewal rate said 84 percent throughout.

Reading it

Pattern Reading
Retention fell for one cohort That year's institutions were sold something that did not fit
Expansion stalled from one cohort on Cross-sell motion or product breadth changed then
Both fine but revenue per institution fell Price or discounting changed that year

Where it goes wrong

Cohorts by contract year, not first purchase. Institutions move between them and the identity fails.

Merged records with two first dates. One institution in two cohorts.

Young cohorts read as stalled. A cohort in its first year has no expansion yet. Mark too young.

Renewal rate as the retention measure. Averages every cohort together.

Every year, by cohort

Mapped once, the orders and the institution master produce the cohort table, the expansion ratios and the identity every year, and the stalled cohort is visible the year it stalls. Covirage builds this from the exports as they are. The education page describes the setup, and the licence utilisation guide covers the usage signal that usually explains a stalled cohort.

Questions people ask

Why by first-purchase year?

Because institutions that joined in the same year met the same onboarding, the same product version and the same sales approach. A cohort that behaves differently from its neighbours is evidence about that year. The renewal rate averages every cohort together and cannot say which year.

How is expansion measured per cohort?

Programmes per institution now, against programmes per institution at the end of the cohort's first year, among the institutions still buying. A cohort at 1.2 programmes in year one and 2.8 now expanded; one at 1.1 and 1.2 did not.

What is the identity?

Every current institution is in exactly one cohort, and the cohorts' current revenue sums to the base. An institution with two first-purchase dates, from a merged record, fails it and is listed.