A cohort funnel follows one defined entry population through specified qualifying stages and observation windows. Keep identities and stage rules consistent, count each member according to the chosen progression rule, and compare equally mature cohorts. Period activity totals can complement the funnel but should not be substituted for its conversion denominators.
Choose the shared entry population
Define the event that admits a member, such as a first eligible enquiry or a new sales opportunity. Specify identity, deduplication and re-entry rules. A person who submits three forms can represent one customer journey or three opportunities depending on the question. Make that choice before counting stages. Keep tests, spam and ineligible entries visible in an exclusion bridge rather than deleting them without a record.
Freeze or version the cohort membership after an agreed arrival window. If a late entry appears, record how it changes the denominator and previously published rates. A shared cohort means later stage events are linked back to that population, even when they occur in later calendar months. Dividing this month's purchases by this month's enquiries is not automatically a journey conversion rate when customers take time to decide.
Define stage order, eligibility and elapsed time
Write the qualifying event for every stage and whether progression must be ordered. A strict funnel may require enquiry, then qualified opportunity, then purchase; a looser path analysis may allow customers to skip a stage. State how reopened opportunities, repeated demonstrations and cancelled outcomes are treated. A member can be counted once per stage even if they trigger the stage event several times.
Use an explicit time window from entry or between stages. The denominators differ: purchase divided by all entrants is an overall rate, while purchase divided by qualified members is a stage-to-stage rate. Label both clearly. Retain the underlying counts so a reader can verify that numerators belong to the stated denominators and understand cases excluded because their observation time is incomplete.
Compare maturity and diagnose drop-off carefully
Compare cohorts at the same elapsed age, and mark recent outcomes provisional. Show stage counts as well as percentages, because a small denominator can produce dramatic movements. A lower stage rate can reflect changed eligibility or customer mix as well as operational performance. Inspect linked records and source coverage before attributing drop-off to a team or a product change.
The analytics lifetime reference provides background on entry and later behaviour. The following funnel is fictional, with its own counting rule. Preserve period activity views separately for staffing and workload decisions. They can show how many demonstrations occurred this month without implying those demonstrations all belong to this month's new entrants.
For members who move backward or repeat stages, retain the event history. Reducing the source to only the latest status can erase the fact that a customer previously qualified, making historical progression impossible to reconstruct reliably.
Retain an explicit reason for a stage failure when the source supports it. Ineligible, not yet observed and observed without progression are different states; combining them into one drop-off category can direct follow-up toward the wrong problem.
Fictional enquiries: April purchases include older journeys
A fictional scenario to explain the method, not a customer result or a market benchmark.
A fictional company receives 100 eligible enquiries in April. Within sixty days of each entry, 40 become qualified and 20 purchase under a strict ordered funnel. Overall purchase conversion is 20 divided by 100, or 20%, while qualified-to-purchase conversion is 20 divided by 40, or 50%.
The company also records 30 purchases during April. Eighteen belong to earlier enquiry cohorts and twelve belong to April entrants. Dividing 30 by April's 100 enquiries gives 30%, but that mixes populations and is not the defined April-cohort outcome. The full sixty-day cohort result remains 20%, observed after maturity.
The report displays both journey conversion and April purchase activity with distinct labels. These invented counts demonstrate denominator and timing alignment. They do not establish why customers converted, prove a causal improvement or imply that a sixty-day window is suitable for every buying process.
TAKE THIS INTO YOUR NEXT CONVERSATION
A practical next-step checklist.
- Define entry event and deduplicated identity.
- Specify stage order and repeat-event rules.
- Label overall and stage-to-stage denominators.
- Compare equally mature cohorts with linked counts.
Common questions.
Can a customer skip a funnel stage?
Only if the definition allows it. State whether the funnel requires ordered progression or measures looser stage participation.
Why do monthly purchases not equal cohort purchases?
Monthly activity includes customers who entered earlier, while a cohort follows the same entrants across later periods.
Further reading & source context.
These primary references provide background. They are not evidence for the fictional example above and do not endorse Accuralytix.
Found an error or a missing qualification? Send an editorial correction.
Check my data readiness.
Use a practical tool to turn this idea into a clearer starting point.
Check my data readiness