Direct answer: create one Metric Dictionary before entering numbers, then add one weekly Creator Metrics Tracker row containing the period, source timestamp, exact values, denominators, missing-value state, operational change note and one Decision Record. Compare like with like; never silently change a definition because a newer dashboard view looks more useful.
This template is for creators who can see numbers but cannot reliably explain what changed between one review and the next. It keeps a small, repeatable record. It does not replace the broader OnlyFans analytics guide, review third-party tools, provide market averages or claim that one metric caused another.
The useful output is not a crowded dashboard. It is a traceable weekly row and a decision that can be revisited. If the source is incomplete, the tracker should say incomplete rather than filling the gap with memory or an estimate.
Freeze the tracking period first
Choose a weekly boundary that can be repeated: start date, end date, time zone and snapshot time. A Sunday-to-Saturday period recorded every Monday morning is one possible local rule. The particular day matters less than using the same boundary whenever practical.
Record whether the source reports events by local time, account time or another platform convention. A late-night post can move between periods when time zones differ. That does not make the platform wrong; it means the tracker needs a declared basis.
Keep partial periods separate. A four-day launch week should not appear beside seven-day periods without a visible coverage note. Label it partial and resist annualising it. The template exists to preserve what was observed, not make every row look comparable.
Build the Metric Dictionary before entering values
The Metric Dictionary is the control sheet. Give every input a metric ID, display name, plain-language definition, source screen or export, numerator, denominator, unit, scope, snapshot timing, allowed missing states and owner. Add the date on which that definition became active.
For example, “profile conversion” is unusable until the creator states which visitor count and which subscriber count belong in the calculation. The separate profile conversion rate calculator owns that formula. In this tracker, store its approved output, source IDs and version rather than inventing a second definition.
Definitions should describe observable fields. Avoid names such as “good fans,” “strong engagement” or “viral content” unless those labels have explicit rules. If a platform renames a field, preserve the old definition and open a new version instead of rewriting history.
Select a compact weekly input set
Start with inputs connected to the creator's current questions. A practical set may cover destination visits, new paid subscribers, active subscribers at period end, expired subscriptions, gross platform-reported earnings, content published and promotion hours. Use only fields the creator can source consistently.
Separate counts from calculated rates. Store the numerator and denominator beside a rate so the result can be checked later. A rate based on eight observations and the same rate based on eight hundred observations carry different uncertainty even when their displayed percentages match.
Do not collect a metric simply because another creator mentioned it. Every field adds capture work and another opportunity for definition drift. Add a field when it informs a recurring decision; archive it when nobody can name the decision it supports.
Record sources and missing values explicitly
Each weekly value needs a source locator: platform area, export name, report date, screenshot ID or stored file ID. The tracker should point to evidence without copying private fan-level information into a broadly shared sheet.
Use named missing states such as not available, not captured, delayed, definition changed and not applicable. Zero is a real observed value; it is not a substitute for missing. A blank cell should never force the next reviewer to guess which state applies.
When a report is still processing, keep the initial timestamp and add a later revision with a reason. Do not overwrite a previously reviewed number silently. A simple revision column protects the connection between the evidence seen and the decision made at that time.
Copy the Creator Metrics Tracker row
The practical artifact is one weekly row plus linked notes. Copy these fields into a sheet or database:
- Period: start, end, time zone, coverage days and snapshot timestamp.
- Metric values: metric ID, value, unit, numerator, denominator and dictionary version.
- Evidence: source locator, capture owner, revision state and quality note.
- Context: content count, promotion hours, known interruption and material account change.
- Decision Record: observation, question, chosen action, owner, due date and next review.
Keep context factual. “Published two fewer feed posts because one shoot moved” is useful. “The algorithm hated us” is an explanation without evidence. Explanations belong in a hypothesis field and should remain distinguishable from recorded events.
Calculate change without hiding the base
Show current value, comparison value, absolute change and percentage change only when the denominator is meaningful. If the comparison value is zero, mark percentage change not meaningful rather than displaying an infinite increase.
Choose the comparison before reviewing the result. Previous complete week, four-week median or same campaign phase can answer different questions. Record which basis was used. Do not swap to whichever comparison makes the current value look strongest.
Round for reading only after preserving the source value. If the platform displays an already rounded number, mark that limitation. Apparent movement smaller than the reporting precision should not become the headline conclusion.
Add a dated operational change note
Record material changes that could alter interpretation: a price change, promotion pause, content-series launch, unusual outage, time away, tracking-link change or major profile revision. Include effective date and the part of the period affected.
This note does not prove that the change produced the result. It prevents future reviewers from assuming the operating conditions were identical. Several overlapping changes reduce what can be learned, and the row should say so plainly.
If the current question concerns subscriber loss, hand the evidence to the subscriber churn diagnosis worksheet. This tracker preserves inputs; that worksheet structures the deeper diagnosis.
Use a fixed weekly review rule
Review completeness before performance. Confirm the period, dictionary version, sources, denominators and revision states. Next, describe no more than three material observations in neutral language. Then select one question that deserves an action or further evidence.
A Decision Record must state one of four outcomes: continue unchanged, investigate, run a bounded test or stop an activity. Name the owner, evidence required and review date. “Post more” is not a decision because it has no quantity, scope or completion condition.
Defer extra ideas to a backlog. A weekly review that launches five changes destroys the ability to read the following row. One documented decision creates a clearer learning trail than a long collection of simultaneous reactions.
Work through a small example
Suppose destination visits are complete, new subscriber count is complete and promotion hours are missing for two days. The creator can calculate the approved destination conversion rate because its numerator and denominator are present, but cannot fairly compare conversion efficiency per promotion hour.
The observation might be: destination visits rose while the conversion rate stayed within the creator's recent range. The tracker does not say why. The Decision Record could continue the current destination for one more complete period while restoring the missing time capture.
At the next review, the creator can compare two complete periods using the same field definitions. The value of the tracker is the preserved boundary, not whether the number moved up.
Maintain the tracker without rewriting history
Review the Metric Dictionary monthly or whenever a source field changes. Version modified definitions, document migration rules and decide whether older rows remain comparable. Do not backfill unavailable history with inferred figures.
Archive unused fields with a reason and final capture date. Keep the original evidence locator accessible under the creator's own retention and access rules. Export a periodic copy so the operating record does not depend on one dashboard view.
Quality-check a sample by returning from the weekly row to the source. Confirm the value, unit, period and denominator. A tracker that cannot be traced is a decorative summary, not a reliable decision tool.
Audit calculated fields separately
Once a month, choose every calculated metric used in a decision and rebuild it from the stored source values. Confirm the numerator, denominator, rounding rule, missing-value treatment and dictionary version. Mark the audit date and reviewer beside the formula rather than assuming a spreadsheet calculation remains correct forever.
If a formula changes, open a new version and identify the first affected period. Recalculate history only when the original inputs remain available and the comparison question requires it. Keep the old result accessible so previous Decision Records still point to the value actually reviewed at the time.
Limitations and sources
Limitations: the tracker describes the creator's selected sources and definitions. Platform reports may be delayed, rounded or revised, and different reporting surfaces can use different scopes. Weekly movement does not establish causation, predict future results or provide a valid comparison with other creators.
Google Analytics documents that reports support monitoring and investigation while also noting differences between reporting surfaces, filters and data processing. GOV.UK performance guidance supports defining measurements around decisions. Those principles inform the source, scope and review fields here; the Creator Metrics Tracker, Metric Dictionary and Decision Record are original SirenCY editorial artifacts.