Understanding Outlier Scores

Outlier scores tell you exactly how much a video outperformed (or underperformed) its channel's typical content.

The Formula#

Outlier Score = (Video Views / Channel Median Views) - 1

Example: A channel's median is 20,000 views. A video with 50,000 views scores:

(50,000 / 20,000) - 1 = 2.5 - 1 = 1.5

This video has a 1.5 outlier score—it got 150% more views than the channel's typical video.

Score Examples#

Score Ranges and What They Mean#

Negative Scores (Below Median)#

• -0.5 to 0: Underperformers—the channel's weaker content
• Useful for understanding what doesn't work

0.0 to 0.3 (Average)#

• Normal variation in video performance
• Not significant enough to study

0.3 to 0.5 (Mild Outlier)#

• Notably better than average
• Worth bookmarking for potential patterns

0.5 to 1.0 (Strong Outlier)#

• Clear outperformance—50-100% above median
• Likely has replicable elements

1.0 to 2.0 (Very Strong Outlier)#

• 2-3x typical performance
• Prime candidates for deep analysis

Above 2.0 (Extreme Outlier)#

• Often due to external factors
• Filtered in some views to focus on reproducible success

Why We Use Total Views (Not View Velocity)#

Some tools measure "view velocity" (views per day since upload). We use total views instead because:

Velocity has recency bias. A video uploaded yesterday with 1,000 views has high velocity, but we can't know if it will sustain that growth.

Total views are stable. After a video matures (30+ days), total views accurately reflect its long-term performance.

Industry standard. Tools like vidIQ and 1of10 also use total views for outlier detection.

This means newer videos may not have accurate scores yet—they need time to mature.

How Median Is Calculated#

For each channel, we:

1. Collect view counts from recent public videos
2. Sort them from lowest to highest
3. Find the middle value (median)
4. Use this as the baseline for scoring

Channels with fewer than 10 videos don't have enough data for reliable scoring.

Comparing Across Channels#

Outlier scores are comparable across channels:

• A 1.0 score on a 5K subscriber channel and a 500K subscriber channel both mean "2x their typical performance"
• This lets you find winning content patterns regardless of channel size

However, absolute view counts differ significantly—focus on the pattern, not the raw numbers.

---

Related Articles

• What is an Outlier Video?
• Finding Outliers From Channels Your Size
• How AI Ideas Work