How to read a YouTube audience retention graph

The audience retention graph is the most useful chart YouTube gives you, and the easiest one to misread. It shows, moment by moment, how many of your viewers were still watching. Read well, it points at the exact seconds where your video lost people. Read badly, it sends you fixing the wrong thing.

This guide covers what the graph shows, the shapes worth recognising, the mistakes to avoid, and a routine you can run on any video in a few minutes.

Where to find it

Open YouTube Studio, choose a video, and go to its analytics. The audience retention graph sits alongside average view duration and average percentage viewed, together with a set of key moments that pick out notable parts of the curve, such as how many viewers stayed through the intro.

What the axes mean

• The horizontal axis is the video's timeline, from the first second to the last.
• The vertical axis is the share of views still watching at that point. If the line reads 60% at 2:00, then 60% of views were still playing at the two minute mark.

The line can briefly rise above 100% where a lot of viewers rewatch a section, because those moments are counted each time they are played.

The six shapes to look for

Almost every retention graph is built from the same few shapes. Learning to spot them is most of the skill.

1. The opening drop

Every video loses viewers in its first seconds, as people who clicked by accident or did not find what they expected move on. What matters is how steep the drop is and how soon the line levels out. A long, steep slide through the first 30 seconds is the most common and most expensive problem on YouTube, which is why why viewers leave in the first 30 seconds gets its own article.

2. The gradual slope

After the opening, most curves decline slowly and steadily. This is normal. A gentle slope means the video is holding people, with a few leaving along the way as they would from anything.

3. Cliffs

A cliff is a short stretch where the line falls much more steeply than the slope around it. Cliffs are the most actionable thing on the graph, because they usually trace back to one moment: a recap, a tangent, a long static shot, a change of topic, or a section the title did not promise.

4. Flat stretches

A flat line means almost no one is leaving. These are the parts of the video that are working, and they are worth studying as closely as the problems. Whatever you did there is something to repeat.

5. Spikes

A spike means viewers rewatched a moment or skipped ahead to it. Both are worth checking. A rewatch often marks something surprising, funny or hard to follow. A skip-ahead often means the part before the spike was something viewers wanted to get past.

6. The end drop

Most curves fall away in the final seconds as the video wraps up and the end screen appears. A long, slow fade through a drawn-out outro is worth tightening. A sharp drop right at the end is normal.

Common misreadings

• Treating every spike as good news. If viewers skip to a moment, the section before it may be the real problem.
• Comparing raw percentages across different lengths. A long video will nearly always sit lower than a short one. Compare like with like, as covered in what is a good audience retention rate.
• Reading meaning into noise. On a video with only a few hundred views, small wobbles in the line are mostly chance.
• Looking only at the average. Two videos can share the same average percentage viewed and fail in completely different places. The shape tells you where.

A five minute routine for any video

1. Check the first 30 seconds. How steep is the opening drop, and when does it level out?
2. Find the two steepest cliffs after the opening.
3. Watch the 15 seconds before each cliff with the graph open beside the video. Note what changes: the topic, the pace, the visuals, the sound.
4. Find the longest flat stretch and note what is working there.
5. Write down one change to make in your next upload, and one thing to keep.

One change per video is enough. It is easier to see what a single change did than to untangle five.

Letting Viewlio do the lining up

The slow part of that routine is step three: lining the curve up against the footage and working out what caused each drop. Viewlio does that for you. Each video report marks the hook, the drop-offs, the gains, the holds and the pacing, then explains what was happening at each moment, reading your title, thumbnail and metadata alongside the curve. A deep dive adds the footage, audio, transcript and on screen text around every event. Tips you want to keep go onto one checklist for your next edit.

You can try it on the free tier with your own videos.

Questions

Why does my retention graph go above 100%?

Because rewatched moments are counted each time they are played. If many viewers replay a section, the line can rise above the number of views at that point.

Why does every video lose so many viewers at the start?

Some viewers click and quickly decide the video is not what they wanted. Every video has this. The aim is a shorter, shallower opening drop, which usually comes from an opening that delivers on the title straight away.

What is a good shape for a retention graph?

A shallow opening drop that levels out quickly, a gentle slope with long flat stretches, and no sudden cliffs in the middle.