Attribution
Part of Gaming advertising experiments
Recording a campaign result when measurement coverage is limited
Document a gaming experiment result with its measured population, unknown delivery and limits on the decision it supports.
Record what the campaign measured and put the gap beside the result. A measured subset is not automatically the whole Australian campaign. Unknown exposure should remain unknown so the next buyer can see which conclusion the evidence supports.
Preserve the intended comparison
Start the result record with the assigned groups, primary outcome, dates, approved games, formats, Australian eligibility rule and decision threshold agreed before launch.
Then record what actually ran. If the planned comparison did not run, report a delivery or feasibility finding rather than presenting the original experiment as complete.
Map the missing data
For each material game, device, format and test arm, record counted delivery and whether exposure and the primary outcome could be measured. Distinguish an event that did not occur from one the system could not observe.
Some metrics available on mobile may not be readily available on PC and console. The campaign report must establish its actual coverage.
Record / Why it matters
- Assigned units and counted delivery by arm
- Shows whether the planned comparison received delivery.
- Measured population and denominator
- Defines whom each reported outcome describes.
- Unclassified game, device or geography
- Shows where eligibility or placement is unknown.
- Missing exposure or outcome by arm
- Shows whether the gaps could affect the comparison unevenly.
- Outcome window and report date
- Shows whether later events could still arrive.
For served game-world ads, report only viewability classifications supported by the applicable measurement method. Separate extrapolated counts from directly identified counts. Do not apply those viewability categories to another format without an applicable method.
Planned vs Actual Campaign Delivery and Measurement Coverage
- Unclassified Devices
- PC and console placements not measurable
- Missing Exposure by Arm
- Arm C: no viewability data; Arm D: partial outcome measurement
Key Measurement Gaps in Australian Gaming Campaign
- Unmeasured Platforms
- Console and PC (no valid measurement method)
- Viewability Classification
- Only mobile-supported methods applied
State a conclusion the record supports
A useful result note names the measured population, outcome definition, period and coverage exclusions in ordinary sentences.
If coverage differs materially between arms, show the measured results but explain why a direct comparison may be misleading. Do not apply a measured rate to unmeasured games or devices without a supported method.
Close the record with the next decision and its reason: use the bounded finding, repeat with better coverage, or treat the outcome as inconclusive. Keep the original report, definitions and corrections together. That makes the result reusable without assigning an outcome to delivery the measurement did not observe.
Steps to Record a Valid Campaign Result with Limited Coverage
- Define the measured population and outcomeInclude dates, eligible games, formats, and eligibility rules
- Document actual delivery vs planned assignmentReport gaps in units, devices, or geographies
- Distinguish non-occurrence from unobserved eventsDo not assume absence where measurement was not possible
- State conclusion based on available evidenceAvoid applying results to unmeasured platforms without support
- Close with next decision and rationaleUse bounded finding, repeat with better coverage, or declare inconclusive
