The verdict below is produced by a fixed set of rules, not a model. It looks at the metric definition, the denominator, the cohort, the statistic, the segment, the dataset and the publication dates, and reports every reason it found.
Adjust 2024 10%↔Adjust 2023 9%
Roughly comparableThe same thing is being measured, but at least one condition differs. Read the direction, not the gap.
- Different sub-segments: not broken out vs calendar year 2022.
- Different aggregation methods. Adjust 2024: medians across verticals, averages for the vertical cuts. Adjust 2023: medians across verticals.
Adjust 2024 11%↔Adjust 2023 9%
Roughly comparableThe same thing is being measured, but at least one condition differs. Read the direction, not the gap.
- Different platforms: iOS vs cross-platform.
- Different sub-segments: not broken out vs calendar year 2022.
- Different aggregation methods. Adjust 2024: medians across verticals, averages for the vertical cuts. Adjust 2023: medians across verticals.
Adjust 2024 8%↔Adjust 2023 9%
Roughly comparableThe same thing is being measured, but at least one condition differs. Read the direction, not the gap.
- Different platforms: Android vs cross-platform.
- Different sub-segments: not broken out vs calendar year 2022.
- Different aggregation methods. Adjust 2024: medians across verticals, averages for the vertical cuts. Adjust 2023: medians across verticals.
Adjust 2024 10%↔Adjust 2024 11%
Roughly comparableThe same thing is being measured, but at least one condition differs. Read the direction, not the gap.
- Different platforms: cross-platform vs iOS.
Adjust 2024 10%↔Adjust 2024 8%
Roughly comparableThe same thing is being measured, but at least one condition differs. Read the direction, not the gap.
- Different platforms: cross-platform vs Android.
Adjust 2024 11%↔Adjust 2024 8%
Roughly comparableThe same thing is being measured, but at least one condition differs. Read the direction, not the gap.
- Different platforms: iOS vs Android.