Direct answer: score one existing OnlyFans offer across four separate dimensions: observable demand, delivery reliability, subscriber-experience evidence and creator workload. Define each target before reviewing the period, show raw counts beside scores, grade source confidence, and choose Keep, Refine one variable, Pause or Investigate. Do not turn the total into a universal benchmark or earnings forecast.
This scorecard reviews an offer already in use. It does not invent a new offer, set its price, forecast earnings or tell every creator what “good” performance looks like. The targets come from the creator's promise, capacity and evidence.
Use the offer ladder worksheet to define an offer's role before scoring it. A score cannot repair an offer that has no clear audience, promise or place in the creator's experience.
Lock the offer and review period
Create an Offer ID and version. Record the included experience, exclusions, audience, displayed conditions, fulfilment steps and current owner. A changed deliverable or promise creates a new version even when the public name stays the same.
Set review start, review end, observation cutoff and time zone. Use one consistent accounting and workload definition. Preserve previous periods rather than overwriting a rolling total.
Name the decision this review must support: keep unchanged, refine one variable, pause new availability or investigate missing evidence. A general “see how it is doing” review invites selective interpretation.
Define creator-owned targets before scoring
Write a target and tolerance for every scored dimension before viewing the final result. Targets might concern eligible demand events, fulfilment completed as promised, unresolved experience issues and minutes per completed delivery. They are operating choices, not market standards.
Use zero, one or two points per dimension: zero misses the target and tolerance, one falls within the pre-written tolerance, and two meets the target. If the required source is unavailable, mark the dimension unavailable and reduce the possible score rather than awarding zero.
Do not change a target because the period produced an uncomfortable number. Change it prospectively in a new scorecard version with a reason.
Dimension one: observable demand
Choose an eligible audience denominator and one demand event that can be supported: qualified request, offer selection, unlock or completed purchase, depending on the offer. Show eligible count, event count and source. Do not combine likes, questions and purchases into one demand total.
Segment only when the segment was defined before review or is clearly labelled exploratory. A small high-intent subgroup can explain a future test, but it should not replace the primary denominator after the fact.
If the offer originates from a documented creator angle, link it to the niche-to-offer map. Niche fit is context, not a score multiplier.
Dimension two: delivery reliability
Count accepted orders or eligible fulfilments, completed as promised, completed late, revised, cancelled and still unresolved at cutoff. Define “as promised” from the offer record, not from a general assumption.
Calculate completion fraction from completed-as-promised divided by eligible fulfilments. Show the raw fraction. An unavailable fulfilment record makes this dimension unavailable; revenue cannot substitute for delivery evidence.
Record recurring failure points such as unclear intake, missing inputs, editing bottlenecks or response backlog. Keep individual details out of the scorecard; store only the operational category and source reference needed for review.
Dimension three: subscriber-experience evidence
Use direct, relevant evidence linked to this offer version: clarification requests, completion acknowledgements, unresolved issues, voluntary feedback or repeat requests. Separate positive, neutral, negative and unclear observations.
Do not treat silence as satisfaction. Also do not let one intense complaint define the whole denominator. Report counts, eligibility and collection method so selective response remains visible.
Create an issue threshold tied to the creator's promise, such as zero unresolved high-impact delivery failures at cutoff. The scorecard supplies the field, not a universal acceptable complaint rate.
Dimension four: creator workload
Count intake, preparation, production, review, delivery, revision, subscriber communication and reporting time. Distinguish fixed setup from per-delivery effort. Use actual minutes where available and flag estimates.
Calculate total minutes per completed delivery only when both values are valid. If completion is zero, show not calculable. Compare with the pre-written workload target and capacity for the same offer version.
Use the PPV break-even calculator for a separate cost-coverage scenario. This scorecard describes observed workload and demand; it does not recommend a price.
Grade source confidence before reading points
Use direct when the offer version, event, fulfilment and window align. Use directional when evidence is relevant but attribution, denominator or completeness is limited. Use unavailable when a required source cannot support the field.
Confidence is a decision cap. Direct evidence can support Keep or Refine. Directional evidence caps the decision at Refine a measurement or Investigate. Unavailable core evidence requires Investigate, regardless of the visible point total.
Check current platform boundaries using OnlyFans Terms and current account records. Do not preserve a static platform-feature claim as an operating assumption.
Copy the offer performance scorecard workflow
Artifact: copy these fields: Scorecard ID | Offer ID | version | audience | promise | review start | review end | cutoff | demand target | eligible audience | demand events | delivery target | eligible fulfilments | completed as promised | late | unresolved | experience target | positive observations | negative observations | unclear observations | workload target | total creator minutes | completed deliveries | minutes per completion | confidence | available points | earned points | decision | one-variable refinement | owner | next review.
Add four evidence rows: dimension | source | numerator | denominator | observed value | target | tolerance | points | confidence note. Then add a decision-cap row so the total cannot override weak evidence.
A fictional example can earn six of eight available points with direct evidence, or six of six available points when one dimension is unavailable. Those results are not equivalent. The second requires Investigate because a perfect-looking percentage can hide a missing dimension.
Convert the score into a bounded decision
Keep when creator-defined targets are met, experience evidence is acceptable and confidence is direct. Refine one variable when a specific dimension has a supported weakness. Pause when unresolved delivery or capacity makes continued availability inconsistent with the promise. Investigate when core evidence is missing or contradictory.
Do not use fixed point bands across all creators. Write the score-to-decision rule before review and retain it with the version. The evidence cap always overrides the arithmetic.
Open a new Offer version when the promise, deliverable, conditions or audience changes. Open a new Scorecard ID for the next period. Link both records so improvement claims compare like with like.
Audit the review for information gain
Ask whether the review changed a decision or exposed a missing source. If every offer receives the same conclusion despite different evidence, the targets or dimensions may be decorative.
Sample source records for each dimension. Confirm the offer version, window, denominator and workload rule. Preserve contrary evidence in the notes.
The UK Government Service Manual recommends choosing measures that reflect the service purpose; see its official guidance on setting performance metrics. The four dimensions, confidence cap and copyable scorecard are original SirenCY editorial tools.
Compare offer versions without erasing the promise
Create a version-comparison view with Offer ID, version, promise change, audience change, review window, available dimensions, raw counts and decision. Only compare a dimension when its definition and source remain aligned. A higher total after the promise changed does not isolate the effect of one variable.
Keep the previous promise visible. Otherwise a score can improve because the creator quietly reduced what was offered, not because delivery became more reliable. Report promise changes beside the observed workload and experience movement.
When the new version serves a different audience or fulfils a different job, treat it as a separate offer for decision purposes even if branding remains similar.
Resolve conflicting dimension signals
High observed demand with weak delivery reliability does not justify Keep. Strong delivery with little supported demand does not automatically justify Pause when the audience denominator is unreliable. Read the dimensions as separate constraints before the total.
Use the decision cap and name the controlling dimension. A note such as “Pause new availability because unresolved fulfilments exceed the pre-written promise threshold” is clearer than “score too low.”
When two dimensions conflict but no gate is controlling, choose one bounded investigation. Collecting more of every metric adds workload without guaranteeing better evidence.
Review scoring integrity
Have a second reviewer recalculate a sample from raw counts. Confirm targets and tolerances were versioned before the result, unavailable fields reduced available points, and confidence was applied after arithmetic.
Track manual overrides with previous decision, new decision, reason, owner and timestamp. Overrides can be appropriate when the creator changes direction, but they should not masquerade as outcomes produced by the scorecard.
Audit whether the scorecard causes action. If repeated reviews produce no decision, remove nonessential fields or make the next-action owner explicit. The tool should improve a creator's operating choice, not create reporting theatre.
Limitations
Limitations: subscriber response is selective, platform records can be incomplete, workloads vary and one review period may be too small for a stable comparison. The scorecard supports a local operating decision; it does not prove market demand, predict earnings or establish a universal offer benchmark.
Revisit targets when capacity or the creator's promise changes, but never rewrite historical scorecards under new rules. Transparent changes are more useful than an artificially smooth trend.