“Subscribers left” is an outcome, not a diagnosis. The useful question is where the experience first stopped matching the promise for a defined group, and what evidence the creator actually controls at that point. This worksheet breaks the subscriber lifecycle into observable stages, aligns comparable cohorts, records changes in offer and delivery, and identifies the earliest supported breakpoint. It refuses to fill missing evidence with a story about why people left.
Direct answer: find the earliest lifecycle stage with a supported break
Define the churn event and observation window first. Group subscribers by a compatible start period, offer version and access state. Map each group through promise at entry, first access, early experience, ongoing delivery and renewal approach. At every lifecycle stage, attach creator-controlled evidence such as profile versions, welcome records, delivery cadence, content availability, service interruptions, access or price changes and observed subscriber actions. The earliest material divergence with evidence becomes the diagnostic breakpoint.
This Churn Diagnosis Worksheet does not provide generic retention tactics, predictive software, an exit-survey design or win-back messages. After a breakpoint is supported, route it to the relevant workflow. The subscriber retention strategies guide owns broader responses; this page owns evidence and diagnosis.
Define churn, cohort and window before analysis
Write the exact event counted as churn. It may be an expired subscription, a disabled rebill that later expires, or another platform-labelled end state. Keep those events separate if their meanings differ. Do not combine an intent signal with a completed lapse and call both churn.
Define the denominator: subscribers eligible to reach the relevant renewal point during the window. Exclude people who have not yet had enough time to experience that point. Record whether trials, free access, discounts, refunds or gifted access are included. Different entry states can create different lifecycle conditions.
Close the observation window after a documented reporting delay. Preserve start and end timestamps, timezone, export date and source labels. A rate with no cohort eligibility rule can change merely because newer subscribers have not yet reached the same stage.
Map the subscriber lifecycle into observable stages
Use five stages. Entry promise is what the public profile and offer communicated at subscription. First access is what became visible immediately after entry. Early experience covers the first defined delivery and interaction window. Ongoing delivery covers the repeated content and service pattern. Renewal approach covers the page state and experience visible before the subscription end point.
Define each stage with dates or events appropriate to the creator’s operating model. Do not rely on “first month” if some subscriptions have different terms. The stage labels must point to records another reviewer can retrieve: page captures, posts, availability logs, support records or platform event exports.
Add post-lapse as an outcome record, not a place to invent a cause. It may contain a directly observed reason if one was voluntarily supplied under a separate research process. Otherwise, leave motive unknown.
Inventory creator-controlled evidence
For entry promise, capture profile, price and offer versions plus the source campaign where known. For first access, record the live welcome, pinned materials, archive availability and any access interruption. For early and ongoing experience, record delivered content dates, stated cadence, content categories, service incidents and the exact availability a subscriber could see.
For renewal approach, record access or price changes, content gaps, unresolved support events and any creator-controlled communication already part of the approved experience. The purpose is not to judge individual conversations. It is to show what the operating system delivered relative to the entry promise.
Tag every item direct, reconstructed or missing. Direct evidence is a preserved export, version or timestamped record. Reconstructed evidence comes from a credible later source but was not captured at the time. Missing means unknown. Never promote reconstructed evidence to direct because it fits the diagnosis.
Build the Subscriber Churn Diagnosis Worksheet
The lifecycle stages, evidence grades and breakpoint rule are original SirenCY methodology. The worksheet does not connect to an account or produce a churn prediction.
| Field | What to record |
|---|---|
| Cohort identity | Entry dates, offer version, access state, eligibility and exclusions. |
| Churn event | Exact platform label, count, denominator, window and close rule. |
| Entry promise | Profile, source and offer evidence available at subscription. |
| First access | Visible welcome, archive, pinned material and access incidents. |
| Early experience | First defined delivery window and observed subscriber actions. |
| Ongoing delivery | Cadence, content availability, service events and changes. |
| Renewal approach | Pre-end page state, price or access changes and unresolved events. |
| Breakpoint | Earliest supported divergence, alternatives, owner and next evidence. |
Keep one sheet per compatible cohort. A summary tab may compare them, but it should link every statement back to the stage evidence and original event definitions.
Compare cohorts without mixing conditions
Compare groups only when they had enough time to reach the same renewal point and share compatible entry state, offer definition, subscription term and event labels. If a price, archive, cadence or acquisition promise changed, either split the cohorts or state that the comparison is descriptive only.
Preserve counts with rates. Report lapses divided by eligible subscribers, not lapses alone. A cohort of 12 with four lapses and a cohort of 120 with 40 lapses share the same arithmetic rate but differ in stability and operational context. The worksheet supplies no universal minimum or benchmark.
Do not select only the best historical cohort as the baseline after viewing outcomes. Define the comparison before stage review: prior compatible cohort, same offer under another period, or no valid comparison. “No valid comparison” is acceptable; the stage map can still reveal a documented break within one cohort.
Locate the earliest supported breakpoint
Read stages in order. At each stage ask: did the creator-controlled experience match the entry promise, did a material change occur, and is the evidence direct enough to support that statement? The first stage with a material, evidenced divergence becomes the candidate breakpoint. Later dissatisfaction may be a consequence rather than the first break.
Require more than coincidence. A cadence gap occurring during a higher-lapse cohort is evidence of changed delivery, not proof it caused each lapse. Write “ongoing delivery is the earliest documented divergence” rather than “subscribers left because posting slowed.”
If every stage matches the records, classify the cause unsupported rather than forcing a creator-controlled explanation. The missing evidence may be visitor expectation, platform conditions, personal subscriber circumstances or something else the worksheet cannot observe.
Write hypotheses and competing explanations
Turn the breakpoint into a testable operational hypothesis: “The cohort encountered a two-week delivery gap that was absent from the approved entry promise; restoring a comparable promised cadence would test whether the same stage divergence remains.” The statement names the evidence and a future comparison without assigning motive to each subscriber.
Add alternatives that the current evidence cannot rule out: different acquisition source, changed offer state, seasonal timing, platform reporting delay or an unrecorded service issue. Rank alternatives by available evidence, not plausibility alone. Mark what new record would strengthen or weaken each one.
Avoid labels such as low-quality subscribers, bored audience or price sensitive unless direct research evidence supports the exact cohort statement. Even then, preserve the original wording and collection method rather than turning a few responses into a universal motive.
Route the diagnosis without prescribing tactics
Assign the breakpoint to the workflow that owns it. Entry-promise mismatch goes to profile or message-match review. First-access failure goes to onboarding or availability operations. Delivery divergence goes to content operations. Measurement failure goes to analytics quality assurance. Include the cohort sheet, evidence grade and competing explanations.
A renewal-stage finding can be checked against the rebill-rate checklist. If direct reasons are needed, use the separate churn exit survey questions workflow. Do not design a survey or send a win-back message from this diagnostic sheet.
Name the next evidence checkpoint and date. The diagnosis closes when an owner accepts the supported breakpoint or records that evidence is insufficient. It should not remain open until every lapse has a personal explanation.
Locate a documented lifecycle breakpoint
Two compatible monthly cohorts entered under the same offer and price state. Entry promise and first access are supported by direct captures. In the later cohort, the planned early experience is delivered, but ongoing delivery contains a documented 12-day gap after an operational interruption. The earlier cohort has no comparable gap.
The later cohort has a higher observed lapse rate at the eligible renewal point. The worksheet names ongoing delivery as the earliest supported divergence and records different acquisition mix as a competing explanation. It does not say the gap caused every lapse. The handoff asks content operations to restore the approved cadence and define a compatible later comparison.
The cohorts, dates and outcome are invented to demonstrate the method. They are not SirenCY results, expected churn levels or evidence that a specific posting gap predicts lapse.
Sources, method and limitations
The lifecycle map, evidence grades and earliest-breakpoint rule are original SirenCY methodology. The platform-model description uses OnlyFans’ response published by Ofcom. Measurement discipline was informed by the UK Government’s performance-data guidance and Google Analytics’ official metrics reference, accessed July 30, 2026.
Limitations: creator records can be incomplete; churn events may lag; cohorts are not randomised; silent subscriber motives remain unknown; and a documented divergence does not prove causation or predict future churn. Use the worksheet to locate a supported operational question, not to classify individual subscribers.
Ready to Scale Your OnlyFans?
See whether there is a genuine fit for strategy, monetization systems, and long-term operational support.
Proposal-specific termsPerformance-basedExit terms documented