OnlyFans marketing KPIs: formulas and a fictional example

Let a fictional weekly acquisition cohort contain 20,000 tracked visitors, 800 new free subscribers and 80 new paid subscribers. Assume for this exercise that the visitors are distinct people, the subscriptions belong to that same cohort and the conversion window is fully observed. Visitor-to-free conversion equals new free subscribers divided by tracked visitors: 800 / 20,000 = 4%. Free-to-paid conversion equals new paid subscribers divided by new free subscribers: 80 / 800 = 10%. Visitor-to-paid conversion is 80 / 20,000 = 0.4%.

Keep the denominator beside the rate

Those assumptions matter. In real reporting, clicks may not identify unique visitors and paid subscriptions may occur after the week ends. A dashboard dividing this week’s new paid subscribers by unrelated current free subscribers would answer a different question. Name the cohort and conversion window in the metric label. Keep raw counts next to rates so a reviewer can see whether a change represents a few people or a substantial population. Do not report a percentage with no denominator or imply person-level tracking the tools do not provide.

Subscriber customer acquisition cost equals included acquisition cost divided by new paid subscribers attributed to that acquisition population. With fictional spend of $1,600 and 80 new paid subscribers, media-only CAC is $20 if the spend represents media alone. A fuller acquisition CAC might also include production, partnerships and management. Do not label the two calculations as interchangeable. Record both when they answer different business questions.

In fictional Week B, 24,000 tracked visitors produce 720 new free subscribers and 72 new paid subscribers with spend of $1,800. Visitor-to-free conversion is 3%; free-to-paid remains 10%; visitor-to-paid is 0.3%; CAC becomes $25. Traffic rose 20% while paid acquisition fell 10%, and CAC rose 25%. That describes the arithmetic, not the cause. Inspect tracking, source quality and offer conditions before concluding why performance changed. An organic source also has a cost in team time and production; no media bill does not mean the acquisition system is free.

Average revenue per user equals revenue for the specified period divided by the specified user population. Always name which users are included: all active subscribers, paying subscribers or a new paid cohort. In this exercise, revenue from the 80 new paid subscribers is $3,600, so paid-cohort ARPU is $45. In Week B, revenue from the 72 new paid subscribers is $2,880, giving paid-cohort ARPU of $40. That is an approximately 11.1% decline in this defined measure.

Total cohort revenue fell 20%, from $3,600 to $2,880. These figures do not establish overall account revenue, profit or retention because those inputs were not supplied. They also do not tell us whether price, purchase mix, refunds or timing changed. Keep gross revenue, platform deductions and net receipts distinct. If refunds are recognized later, explain how the report is revised. A precise name such as “new paid cohort gross revenue per subscriber” is longer than ARPU but prevents several important misunderstandings in a founder discussion.

Subscriber retention over a defined interval equals the number of starting subscribers still active at the end divided by the number active at the start, under your stated activity definition. Churn equals starting subscribers lost during that interval divided by starting subscribers. For a fictional opening cohort of 100 paid subscribers, 80 remaining active gives 80% retention and 20% churn, assuming no reactivation complications and one consistent subscription definition.

Do not include newly acquired subscribers in the ending retained count for that opening cohort. Doing so can hide losses behind growth. Distinguish cancellations, expiration and successful renewal when the data supports those states. Revenue retention is another measure and can move differently from subscriber retention when spending changes. Choose the definition that answers the decision in front of the team. The two-week acquisition dataset above contains no renewal history, so it cannot support a retention or churn calculation. Saying that a metric is unavailable is more accurate than filling a dashboard with a convenient assumption.

Customer lifetime value can mean expected lifetime revenue or expected lifetime contribution margin. Those are different measures. A shortcut sometimes used in a stable subscription model is period ARPU divided by period churn, but it assumes stable economics and churn behavior. It can be badly misleading with immature cohorts, changing offers, irregular purchases or a churn rate based on a different population. Do not present the shortcut as an observed lifetime result.

A stronger approach follows cohort revenue and relevant costs across time, separates observed value from a projected tail and explains the assumptions behind that projection. If only acquisition-week revenue is available, report that limited observation rather than inventing LTV. A comparison of CAC with modeled LTV should use compatible revenue or margin bases and acknowledge uncertainty. It does not automatically establish profitability or a safe scaling budget. The purpose of a model is to make assumptions inspectable and improve a resource decision, not to produce a large number that justifies more acquisition spending.

Use a small table with the counts, rates, period and cost basis needed for the current question. Add a note about missing attribution, conversion lag or tracking changes. Compare cohorts at the same maturity where possible. If the report includes several sources, keep an unknown group and show whether the allocation method could change the recommendation. Avoid adding ratios simply because a spreadsheet can calculate them.

For the fictional weeks, a useful next step is to verify visitor and subscription tracking, inspect source mix and review the offer or purchase composition. These are checks, not proven explanations. Define who owns each check and when the team will revisit the decision. Share aggregate findings with backend managers and chatting leads without exporting private subscriber data. A good KPI system makes it easier to notice uncertainty and coordinate action. It should prevent a founder from mistaking more traffic for better acquisition or early cohort revenue for lifetime value.