YouTube Key Moments for Audience Retention: Intro, Top Moments, Spikes and Dips

The key moments for audience retention report is the part of YouTube Analytics that reads your retention graph for you. YouTube describes it as a report that "explains how well different moments of your video held viewers' attention" and highlights four kinds of moment: intro, top moments, spikes and dips. It lives at the video level only, under the Overview or Engagement tab, and YouTube notes that the data typically takes one to two days to process (YouTube Help).

This article quotes what the report measures, then shows how to carry each finding back into the next script, which is the only place a retention problem is cheap to fix.

The four moments, in YouTube's words

YouTube also describes the shape of the line itself: a flat line means viewers are watching that part "from start to finish", and gradual declines mean "viewers are losing interest over time", which it says is normal, because "videos on YouTube generally taper off during the playback period".

Two limits matter before you read any of this. YouTube notes that moments are "only highlighted if they're detected in a video", and that the video "should also be at least 60 seconds long and have at least 100 views". A new upload with 40 views has no key moments yet, and a 45-second video never will.

Intro: the one number for the first 30 seconds

The Intro figure is the share of viewers still watching after 30 seconds. YouTube gives two readings of a high figure: the first 30 seconds "matched the viewer's expectation of the video's thumbnail and title", and "the content kept the audience interested". Its two recommendations for a low figure are the same pair in reverse: change the thumbnail and title "to better reflect your video content", and "modify the first 30 seconds of your video and experiment with different styles".

That order is worth keeping. A low Intro figure is as often a packaging mismatch as a weak opening. If the title promises one thing and the first lines deliver another, the viewers who leave are not bored, they are correcting a wrong turn. Check the title against the first 90 seconds of the script before you rewrite the hook. The first 30 seconds guide covers the three-phase opening, and the hook examples show 36 openings that state a reason to stay in the first line.

What the script can do about it: in our own 349-video hook study, the openings of videos that flopped were 60% more likely to give no reason to keep watching than the openings of hits, and hits matched their title's promise more often (87% against 78%). Those are the two things the Intro number is reading.

Top moments: move them earlier

YouTube's advice on top moments is direct: "If the top moments are occurring in a later portion in the video, consider introducing the compelling content earlier in the video", because "audience sizes typically decrease over the length of the video". The second suggestion is to make new videos by "expanding on the content from the top moments".

In script terms, a late top moment is a payoff that arrived after most of the audience had gone. Find the passage, then ask whether anything between the opening and that passage was load-bearing. Often it was setup that could be cut or moved after the payoff. The script structure guide covers hook, build and payoff timing.

Spikes: two causes, opposite fixes

A spike is a rewatched or shared moment. YouTube gives both readings: "Your audience watched that segment more than previous segments", or "Your content isn't clear and your audience had to rewatch a section". A spike on a demonstration or a reveal is the first kind. A spike on a dense explanation, a fast list, or a number read once is often the second.

The public version of this signal is the most replayed graph on the watch page, which we read for every transcript page and summarise on Script X-ray. It counts replays, not viewers who stayed, so treat it as a map of what was worth hearing twice, for either reason.

Dips: the passage to fix before the next recording

Dips are skipped or abandoned moments. YouTube suggests you "review your dips to better understand why audiences lost interest in a particular segment". The reasons we see most often when a dip is mapped back onto the script are a sponsor read placed before the first payoff, a repeated point, a digression the title did not promise, or a long stretch with no new information. The retention drop diagnosis guide sorts the curve shapes by cause.

Typical retention and segments

YouTube offers two comparisons beside the graph. "Typical retention" compares the video with "your 10 latest videos of similar length", and the SEE MORE view shows "how your video compares to all YouTube videos of similar length". The segments report splits the curve by new against returning viewers, subscribers against non-subscribers, and organic against paid traffic. A dip that appears only for new viewers is usually a context gap: the script assumed knowledge a first-time viewer did not have.

One metric note from the same page: average view duration is the "average minutes watched among those who stayed to watch", calculated "from engaged views and their corresponding watch time". Viewers who left in the first seconds are not in that average. What is an engaged view explains the distinction.

Carry the finding back into the script

1. •Open the published video's retention report and note the timestamp of each dip and spike.
2. •Convert each timestamp to a position in the script. At the pace you actually spoke, a words to minutes converter turns 4:20 into a word count, and the transcript page for the video shows the words said at that moment.
3. •Write one line next to each dip: what the passage was doing, and what the viewer had been promised.
4. •Fix the pattern, not the passage, in the next draft. A sponsor read before the first payoff will dip again in the next video if it sits in the same place.
5. •Run the next draft through the free script check before recording. It returns a relative attention-risk map over the draft and the biggest flagged passage with its rewrite. It reads text only and does not know how published viewers will behave, which is exactly why you compare it against the dips you already measured.

For a fuller read of the graph itself, see how to read a YouTube retention graph and the audience retention benchmarks.