Direct answer: review a completed seasonal campaign by freezing its planned asset list, recording what actually published, pairing every result with its metric definition and observation window, and adding creator workload beside audience response. Compare only rows with compatible jobs and denominators. Close each asset as Repeat, Modify, Reuse input, Retire or Inconclusive, then carry forward one bounded production decision rather than declaring the whole theme a success or failure.
The loudest number in a seasonal campaign is often the least useful one. A hero post may collect the most visible reactions because it had the largest audience, while a modest preview may have done its assigned job with far less production effort. If the review copies both counts into one ranking, it rewards exposure and ignores purpose.
This tracker starts after the seasonal campaign has ended and its observation windows have closed. It does not plan a campaign, attribute all revenue to one asset, forecast future seasonal interest or supply universal performance thresholds. It reconstructs what happened using the creator’s available records and shows which evidence is comparable.
Freeze the campaign record before interpreting it
Copy the approved campaign plan into a review snapshot. Record the theme spine, campaign dates, intended audience moments, required assets, optional assets, fallback, publish windows and creator-selected goals. Preserve the version that guided production rather than updating it to match the outcome.
Then add actual states: published, substituted, delayed, cancelled or failed verification. A missing asset is operational evidence. Do not remove it from the denominator simply because it never reached the audience.
If the campaign passed through the seasonal readiness checklist, bring its last decision and open risks into the snapshot. That provides context for scope cuts or late repairs without turning readiness into a performance score.
Restate the job of every asset
Give each asset one primary job from the original journey: recognition, anticipation, main delivery, continuity, close or another creator-defined function. A metric matters only in relation to the job it was selected to observe.
A recognition asset might be reviewed for qualified destination visits or profile actions if those records exist. A main delivery asset may use purchase or completion evidence available on the destination. A close asset might be judged by whether it completed the sequence and reduced unresolved production work, not by whether it outperformed the hero.
Do not retrofit the job after seeing the strongest number. If the intended observation was not captured, mark Evidence missing. Changing the question after the result makes any asset appear successful.
Build a metric definition card
For every value, record the source surface, exact metric label, numerator, denominator, start and end times, time zone, extraction time and any filter. “Engagement” is not a definition. “Recorded interactions divided by verified impressions for Asset S-04 during the declared window” is inspectable when both fields exist.
Use the label shown by the current platform rather than assuming metrics share definitions across services. YouTube’s official engagement help, for example, distinguishes views, watch time and average view duration and notes that reporting methods can change. That does not define an OnlyFans metric; it demonstrates why every value needs its own source card.
Preserve raw counts beside calculated rates. If a denominator changes or a row is later found incomplete, the reviewer can recompute the value without reconstructing the export.
Close one observation window per comparison
Choose the window before reading the result where possible. Use the same elapsed duration for assets intended to be compared, or mark the comparison invalid. A post observed for two hours and another observed for seven days should not sit in the same rate column without a clear distinction.
Record late edits, schedule changes, distribution failures and source delays. Do not quietly extend a weak row or close a strong row early. When records process at different times, preserve the extraction timestamp and rerun all affected rows together.
Google Analytics’ official date-range help shows how a reporting tool can compare a selected period with a previous period and can match weekdays in one comparison mode. That is a general example of declaring date logic. It does not make unrelated seasonal periods equivalent or prove causation.
Compare like jobs with compatible exposure
Create comparison groups only when assets share the same primary job, metric definition, denominator type and observation length. Then record meaningful differences such as format, placement, publish window or creative treatment. If several variables changed, call the row Descriptive rather than a test.
Do not rank raw totals from different audience sizes. Use a rate only when the numerator and denominator come from compatible records. If an impression count is unavailable, preserve the count and its exposure limitation rather than estimating a denominator.
Small numbers can move sharply with one action. Label low-evidence comparisons Inconclusive instead of turning a large percentage swing into a campaign lesson. This tracker does not set a minimum sample size because the available records and decision risk vary.
Put creator workload beside response
Record creator-entered capture, selection, edit, caption, review, export, upload and message-support time for each asset or asset group. Add direct production costs only when the creator has reliable records. Do not invent an hourly value or convert effort into revenue attribution.
Note recovery burden, reshoots, dependency failures and last-minute repairs as operational flags. Two assets with similar audience response may deserve different decisions when one repeatedly breaks the creator’s capacity.
Workload is not a punishment score. A demanding hero asset can remain worthwhile when it serves a unique approved job. The point is to make the trade-off visible before the next seasonal package copies the same production shape.
Separate content evidence from seasonal timing
A campaign contains both creative assets and a time context. The results cannot prove that the season caused the response. Existing audience activity, promotion, price, subscriber mix, competing posts and platform behavior may all differ.
Record the actual publish windows and deviations, then use cautious language: “Asset A recorded more of the defined response than Asset B in this campaign” rather than “evening is the best time.” A single seasonal window cannot establish a universal send or post rule.
If timing deserves a test, carry it forward as a separate one-variable experiment with matched assets and declared observation windows. Do not use this retrospective to manufacture a test that was never planned.
Copy the seasonal campaign tracker
Practical artifact: retain raw evidence, compatibility and workload in the same row. The decision field must name what can change next.
| Asset / job | Published evidence | Metric card | Workload | Decision |
|---|---|---|---|---|
| S-01 / recognition | Version, surface, actual time, changes | Raw values, denominator, window, extraction | Creator-entered tasks + repair flags | Repeat, Modify or Inconclusive |
| S-02 / anticipation | Published, substituted, delayed or cancelled | Comparable group or Descriptive only | Time, direct cost if known, recovery note | Reuse input or Retire |
| Campaign close | Planned versus actual package | Missing evidence and contamination | Total load + peak bottleneck | One next production decision |
Use five closeout states
Repeat means the asset job and production shape have enough local evidence to run again without claiming the same result. Modify means one defined component should change. Reuse input means the source asset can feed a future recut or refresh without copying the full campaign.
Retire means the creator chooses not to repeat the asset in its current form because of evidence such as persistent quality failure, unusable workload or no fulfilled campaign job. Inconclusive means the records cannot support a change.
Send eligible source assets to the seasonal reuse planner for Refresh, Recut or Retire decisions. Send a single post that needs deeper job-and-response analysis to the content performance review template. This tracker stays at campaign level.
Run a decision audit before closing
For each recommendation, point to the rows that support it and name at least one plausible alternative explanation. “Shorter edit required less workload and recorded a compatible completion signal” is evidence. “Fans prefer shorter videos” is an audience claim the campaign may not prove.
Check that weak evidence did not vanish from the narrative. Missing denominators, delayed publishes, substituted assets and unrecorded labour should appear in the closeout. An attractive summary that removes these limits is less useful than an honest Inconclusive state.
Choose one next production decision. Examples include keeping the smaller asset package, testing one edit treatment, protecting an earlier review gate or dropping an optional setup. Multiple new experiments make it difficult to learn which change helped.
Worked example: a fictional three-asset campaign
A campaign contained a recognition still, a short anticipation clip and a hero set. The still and clip used compatible observation windows, but the hero set’s destination report used a different denominator. It remains Descriptive and is not ranked with the other two.
The clip recorded the stronger compatible response but also required an unexpected reshoot and caption repair. The decision is Modify: retain its audience job, simplify the capture setup and protect a caption-review checkpoint. The result is not used to claim that clips outperform stills generally.
The hero source contains several approved details suitable for later recuts, so it becomes Reuse input. The campaign close records one next decision: repeat the minimum three-moment journey with the simplified clip process. Revenue attribution and next-season demand remain outside the conclusion.
Official source notes
Google Analytics’ official date-range comparison help, accessed July 30, 2026, documents explicit selected and comparison periods, including a weekday-matched option. It is a reporting-method example, not an OnlyFans measurement instruction.
YouTube’s official engagement help, accessed July 30, 2026, defines platform-specific measures such as views, watch time and average view duration and notes reporting caveats. Those definitions are not transferred to OnlyFans.
Limitations
Limitations: this tracker cannot prove that a season, asset, publish time or creative choice caused the observed result. It cannot forecast the next campaign, subscriber behavior, revenue or trend direction.
Platform labels, counting methods, processing times and available denominators can change. Creator-entered workload may be incomplete. Preserve source labels, extraction times and missing evidence rather than forcing every row into one score.
The review covers one completed seasonal campaign. It does not plan the next package or create universal benchmarks. A Repeat state means repeat the defined local production decision, not expect the same response.