Audience retention — reading the curve and fixing the drops

Audience retention is the percent of viewers still watching at each moment in your video. It's measured at every second of playback. Studio's retention curve is the single most actionable analytics view you have.

How to access it

Open any video → Analytics → Engagement → Key moments for audience retention. The graph shows percent of viewers vs. position in the video.

The y-axis is percent retention (100% = everyone who clicked still watching). The x-axis is video time. The curve always declines because some viewers leave at every moment.

The benchmark line

Studio shows your video's retention against a benchmark for the same video duration and category. This is more useful than absolute retention numbers — a 30% absolute retention on a 20-minute video can be excellent if the benchmark is 20%.

Five patterns to recognize

Pattern 1: Cliff in the first 30 seconds

• Viewer clicked from an interesting thumbnail/title
• The opening didn't deliver
• Often the longest cliff in the entire video

Fix: Rewrite the first 15 seconds. State what the viewer will get and start delivering immediately.

Pattern 2: Steady decline, no big drops

• This is healthy. Some attrition is normal.
• If the slope is steep, the content might not match the audience's expectation set by the thumbnail.

Fix: Tighten editing pace. Cut filler. Re-pace the value delivery.

Pattern 3: Sudden drop in the middle

• Often a topic shift or an ad break
• Could be a slow section that lost the audience

Fix: Find the timestamp. Watch the 30 seconds around it. Identify why viewers leave (slow segment, off-topic, low-energy delivery).

Pattern 4: Spike up (re-engagement)

• Rare and good — viewers go back to re-watch a section
• Usually means a particularly clear or interesting moment

Fix: Make more of these moments. Replicate the structure.

Pattern 5: Long flat run

• Viewers are engaged. Content is delivering.

Fix: This is what you want. Note what's happening in this section and replicate it.

How long should videos be?

There's no single answer, but the pattern is consistent across YouTube:

• 8–15 minutes is the sweet spot for most niches
• Shorter videos must be tightly edited; pacing is everything
• Longer videos (20+ min) need chapter structure and high information density to hold retention

What good retention looks like by length

• Shorts (<60s): 70%+ retention is strong
• 1–3 minutes: 65%+ retention is strong
• 8–10 minutes: 50%+ is strong, 40%+ is acceptable
• 15+ minutes: 35%+ is strong, 25%+ is acceptable

What retention does NOT measure

• Whether viewers liked the content
• Whether they will return
• Whether they subscribed
• Watch-after-pause behavior (those count)
• Background-tab listening (counts as watching for retention purposes — yes, even when tab isn't focused)