Skip to main content

C2. Persistent Demand (by cohorts, not event)

Construct: in the ICP, demand is recurring, reactivatable, or reappears by cohorts over time (it does not depend primarily on a one-off event). Qualifier: persistence must be compatible with the natural cadence of the ICP job/cycle (including long cycles); long cadence does not eliminate the need for a reactivation mechanism.

Application​

In the ICP, the “job” comes back over time (recurring), or can be reactivated, or reappears predictably in new cohorts — without depending mainly on a one-off event spike. The cleanest way to see this is to look at cohort behavior over time, because aggregate averages hide “spike + drop.”

Examples​

Example 1 — Truly recurring (short cadence) Unit: SMEs that issue invoices (NFs) and close monthly books → issuance+reconciliation offer → vs spreadsheet+accountant+government portal.

Works:

  • The work comes back every month (closing/compliance obligations).
  • You can point to a natural recurrence mechanism: “every month has closing/reporting/delivery.”
  • In cohorts, it is not “bought and disappeared”: demand reappears at the same pace as the job.

Does not work (common trap):

  • Demand spikes only when there is a “turning point” (e.g., one-time rule/integration change) and afterward the ICP no longer has a recurring job; it becomes a “project,” not a routine.

Example 2 — Long cadence, but reactivatable (your model’s qualifier) Unit: regulated companies that need to maintain audit/compliance → evidence+controls management offer → vs spreadsheets+consulting+internal tool.

Works:

  • The cycle is annual/semiannual (long), so low frequency is not a problem.
  • But there is a clear reactivation mechanism: periodic audits, renewals, control reviews, team/process changes.
  • In cohorts, you expect to see return/reactivation aligned with the job calendar (not “disappear forever”).

Does not work:

  • “Compliance” was only to pass a one-time audit (e.g., a specific certification that does not repeat in the ICP), and then demand becomes “dead archive” with no reuse/reactivation trigger.

Example 3 — “One-off event disguised as market” Unit: teams that need to migrate once from tool A→B due to deprecation/force majeure → automated migration offer → vs consulting/scripts.

Works (only if by cohorts, not by spike):

  • There is a continuous pipeline of new cohorts facing migration over time (e.g., companies in different contract/stack windows), not a single “big bang.”

Does not work (classic case):

  • Demand depends primarily on a one-off event (a concentrated regulatory/technology change) that creates a boom and then dries up.
  • In cohorts, this appears as “enter, solve, done” with no natural reactivation mechanism. (This is exactly the type of thing cohorts help avoid confusing with sustainable traction.)

Example 4 — Seasonal is OK (as long as it is “natural cadence”) Unit: retailers hiring temporary workers for seasonal dates → recruitment+fast screening offer → vs referrals/agency/spreadsheet.

Works:

  • Demand is not continuous, but reappears by cohorts predictably (e.g., Black Friday/Christmas every year).
  • There is a clear mechanism: seasonal planning + annual replacement + rehiring.

Does not work:

  • You confuse structural seasonality with an extraordinary spike (e.g., “an atypical Black Friday” or “a temporary subsidy”). Normal seasonality returns; extraordinary spikes do not.

Note:​

C2 is about the temporal structure of demand in the ICP. You use cohorts to avoid the mistake of calling an “event spike” persistent demand.