Direct answer: after a completed PPV campaign, first validate its record; then remove prior purchasers and recipients who should not receive a comparable send; check experience duplication and creator capacity; identify the weakest supported campaign stage; and choose Stop, Investigate, Adjust one variable, or Plan a new send. A resend is a new campaign with a new ID, not an extension of the old row.
This tree decides the next action after one completed campaign. It does not set up the original campaign, provide message libraries, recommend a universal resend interval or guarantee a lift. The safest default when evidence is incomplete is Investigate, not “send again.”
Before entering the tree, reconstruct the original plan with the PPV message planner fields: asset, audience, offer, promise, window and planned success signal.
Gate one: validate the completed record
Confirm Campaign ID, asset version, offer, audience rule, exclusions, sent and delivered counts, unlocks, observation cutoff, revenue definition and workload. If a critical field is missing, the result branch is Investigate. Guessing a denominator creates a confident decision from an unstable record.
Check that the evaluation window has closed under the original rule. Do not declare a weak result early and then give a preferred campaign more time. Preserve early and later snapshots when delayed activity matters.
Use direct, directional or unavailable evidence. Only direct evidence can support a clean performance comparison; directional evidence can still identify the next measurement improvement.
Gate two: rebuild the truly eligible audience
Exclude prior purchasers of the same experience using the best available purchase record. Also apply creator-approved exclusions for inactive state, unresolved support, prior duplicate exposure or another relevant condition. Record counts and sources.
Do not treat everyone who did not unlock as eligible for another send. Some did not receive it, some already own an equivalent asset, some left the eligible state and some may have encountered unclear information.
If the remaining audience cannot be identified without guessing, Stop the repeat workflow and repair the audience record first.
Gate three: test meaningful experience difference
Compare the proposed new experience with the original: asset, preview, promise, message purpose, access conditions and audience state. A new caption does not automatically make the same locked asset feel new.
If the intended recipients substantially overlap and the experience would be indistinguishable, choose Stop or design a genuinely different offer. Preserve the old asset for future eligible subscribers rather than assuming immediate repetition is valuable.
If a new audience has not encountered the experience and the asset remains current, continue. Eligibility is a factual gate, not proof that another send will perform.
Gate four: locate the weakest supported stage
Review the chain in order: eligible audience, delivered, opened when available, unlocked, supported revenue and workload. Use raw counts beside fractions. The first weak or unavailable stage guides what should change.
Low delivery or unreliable audience selection points to data or execution, not necessarily the offer. Delivery with little observable attention may support testing presentation. Attention with few unlocks may support an offer or promise hypothesis. Unlocks with excessive support workload may support an experience-clarity review.
These are hypotheses, not diagnoses. Select one stage and write competing explanations before choosing a change.
Gate five: protect creator economics and capacity
Record additional preparation, audience review, response handling and reporting time. Compare it with a creator-selected workload ceiling. A campaign can generate revenue while consuming capacity needed for promised subscriber experiences.
Use the PPV break-even calculator when the decision depends on covering a defined campaign cost. Keep observed campaign performance separate from a forward-looking scenario.
If additional work exceeds capacity or the scenario cannot be supported from current data, Stop or Investigate. Do not treat sunk production effort as a reason to repeat.
Choose one branch and one changed variable
Stop when eligibility is too small or uncertain, duplication is high, the asset no longer fits, workload is unacceptable or the creator chooses not to continue. Investigate when records, delivery or attribution are incomplete. Adjust when evidence supports testing one variable. Plan a new send when eligibility, difference, capacity and test logic are all documented.
Possible variables include audience rule, preview, description, offer structure, message purpose or asset version. Change only one material variable if the goal is learning. If several elements change, label it a new concept rather than a comparison test.
Do not use the offer ladder worksheet to justify an unsupported upsell. Use it only to check whether the proposed offer already has a clear creator-defined role.
Copy the PPV resend decision card
Artifact: copy these fields: Decision ID | original Campaign ID | record complete? | evaluation cutoff | evidence grade | asset version | prior purchasers excluded | equivalent purchasers excluded | remaining eligible count | audience source | experience difference | weakest stage | competing explanation A | competing explanation B | creator minutes | workload ceiling | chosen branch | one changed variable | challenging outcome | new Campaign ID | reviewer | review date.
Run this branch order: incomplete record means Investigate; unverifiable eligibility means Stop and repair; high duplication with no new value means Stop; unsupported weakest stage means Investigate; excessive workload means Stop or simplify; one supported hypothesis with valid eligibility means Adjust and create a new campaign record.
A fictional case might show a complete original row, 18 supported prior purchasers, 47 remaining eligible subscribers, uncertain delivery labels and a valid asset. The correct leaf is Investigate delivery evidence, not a new send. The values illustrate the tree and are not thresholds.
Create a clean handoff for a new campaign
A Plan-new-send outcome should name a new Campaign ID, owner, audience snapshot time, approved change, unchanged conditions, evaluation window and stop condition. Never reopen and overwrite the original campaign.
Record why the nearest alternative branch was rejected. For example, “Adjust preview rather than offer because delivered count is supported and subscriber questions concern what is included.” This makes the reasoning reviewable.
After the new window closes, compare the two records using matching definitions. A better result is still local evidence, not proof that the change always works.
Use live sources and keep fixed rules out
OnlyFans publishes current platform conditions at OnlyFans Terms. Check the live source and current interface before relying on a feature, message state or account rule.
The UK Government Service Manual's performance-metric guidance supports defining measures from a service purpose and using several sources rather than selecting one convenient total. The branch structure, decision card and evidence grades are original SirenCY editorial tools.
Avoid copying timings, purchase-rate claims or revenue examples from search results into the tree. Those conditions rarely match the creator's audience, asset, page or evidence.
Audit decisions against later evidence
When a new campaign closes, return to the earlier decision card. Check whether the audience snapshot, changed variable, workload estimate and evaluation rule were executed as approved. A result cannot validate the tree when the planned branch was not actually followed.
Compare the anticipated weakest stage with the observed stage. If the tree repeatedly points to presentation but later audits reveal audience-data failures, improve the evidence gate rather than adding more creative branches.
Keep stopped decisions in the sample. Reviewing only campaigns that proceeded creates survival bias and hides whether the tree prevented duplicate or poorly supported sends.
Handle disagreement between reviewers
Ask each reviewer to select the first blocking gate independently, then compare paths. If both see the same evidence but choose different leaves, clarify the branch definition. If they rely on different source records, repair the evidence packet.
Record a minority note rather than forcing false certainty. A decision can proceed under the creator's chosen rule while preserving why another reviewer preferred Stop or Investigate.
Escalate only the specific disputed gate. Reopening the entire campaign history every time makes the tool too slow to use and encourages reviewers to skip it.
Track tree health without measuring sales success
Useful operating measures include complete decision cards, prior-purchaser exclusions verified, decisions returned for missing evidence, estimated versus actual workload and new campaigns that preserved one-variable changes. These measures show whether the workflow is being followed.
Do not grade the tree solely by revenue from campaigns it allowed. Its purpose also includes stopping unsupported, duplicative or unmanageable sends. Record prevented work and repaired measurement gaps as separate outcomes.
Review branch frequency after a creator-selected sample. A branch chosen almost every time may reveal unclear definitions or a default habit. Inspect evidence before changing the tree.
Maintain a future-eligible note for a valid asset that is stopped only because the current audience is exhausted. Record the audience state that would make a new review relevant, such as a later cohort that has not received an equivalent experience. This is a review trigger, not an automatic send rule, and it expires when the asset or promise changes.
Separate a new-cohort decision from a repeat-to-the-same-cohort decision in reporting. Both may use the same source asset, but their duplication risk, audience evidence and learning question differ. Combining them can make a broad asset total look like support for immediate repetition when the observed value actually came from newly eligible recipients.
Limitations
Limitations: platform reporting may not expose every audience state, purchaser history can be incomplete, and subscriber choice cannot be reduced to one funnel stage. The tree structures a local decision; it does not establish causality or predict future unlocks.
Review the decision after any material change to the asset, audience source, offer or platform surface. A prior Plan branch does not remain valid indefinitely.