Understanding the junbi benchmark
The junbi benchmark converts neural prediction data into standardized percentile scores by comparing your video ad against a dataset of over 20,000 real-world YouTube campaigns. Every new ad uploaded to junbi is automatically evaluated against this database based on its designated viewer mode and ad type. This standardized baseline helps you measure how effectively your video creative captures and maintains human attention relative to current market standards.
Strategic Value
Raw predictive scores lack context without a standardized reference point. Comparing your creative against thousands of tested video ads provides immediate clarity on whether your visual hooks succeed or fail in capturing audience focus.
How the junbi benchmark work
Every video ad tested in junbi is measured against a dedicated benchmark cohort to ensure an accurate comparison. The system applies benchmarks based on two variables selected during upload:
- Viewer Mode: The environment where the video is displayed, including desktop, desktop theater, mobile, or connected tv.
- Ad Type: The specific YouTube format, including skippable in-stream, non-skippable in-stream, or bumper.
How benchmarks influence scores
Benchmark scores represent relative percentile rankings rather than absolute percentage scores. An 85th percentile score means your creative outperforms 85% of ads in the baseline dataset. Because market standards and media consumption habits shift over time, benchmark datasets reflect evolving visual trends. This ensures your percentile rankings always place your creative accurately relative to the broader ad distribution.
💡 Pro Tip:
Matching your upload configuration to your planned media buy ensures that percentile rankings accurately reflect real viewing conditions.