This page describes a proposed study. It contains no campaign performance findings and is excluded from search indexing.
This page contains the method for a future performance study. It contains no observed campaign benchmark, ROAS claim, or case-study result. Real campaign data and permission to publish it are required before findings can be added.
Download the performance protocol ↓Define the eligible dataset
Include campaigns with documented paid spend, complete dates, known geography, a stated offer, and a consistent business-outcome definition. Record which accounts and periods were eligible and why any were excluded.
Keep organic ChatGPT referrals outside the paid cohort. Preserve different currencies, attribution definitions, and sales-cycle maturity instead of combining them into a single unexplained average.
Measure the business result
Collect media spend, other acquisition costs where included, clicks, confirmed conversions, qualification, revenue, refunds, and contribution as applicable. Record source systems and data freshness. Missing values remain unknown.
Distinguish platform-attributed performance from a causal claim of incremental lift. An observational report cannot establish what would have happened without advertising merely by comparing before and after totals.
Report uncertainty and selection
Publish cohort counts and the distribution of results, not only a selected success story. Explain sample limitations, incomplete follow-up, campaign changes, and the selection process. Avoid presenting a small sample as a universal channel benchmark.
Release a redacted data dictionary and reproducible calculation method alongside the narrative. Obtain permission for any identifiable account examples before publication.
Release only when evidence is ready
The publication checklist requires verified source totals, documented exclusions, consistent currency treatment, reproducible metrics, and a review of claims against the actual data. A result remains a draft until those conditions are met.
Until then, use the budget calculator and test plan for explicit assumptions. They support planning without implying that an unobserved return has already been achieved.