Read the analytics, then double down

Ignore vanity totals. Four metrics diagnose everything:

• Retention / average view duration: did people stay? The single most important number on every platform.
• Swipe-through or hook rate: what share of people who saw the first seconds kept watching? Below the platform norm means the hook fails.
• Click-through rate (YouTube long-form): are people choosing the video from the feed? This is packaging: title plus thumbnail.
• Views from non-followers: is the platform still testing you on new audiences? A healthy shorts channel is mostly non-follower views.

The retention curve tells you exactly where to fix the format:

• A cliff in the first 3 seconds: the hook fails. Rewrite openings.
• A steady steep slope: pacing is too slow everywhere. Cut sentences, speed up.
• A sudden mid-video drop: find that exact moment, it is a boring segment or a broken loop. Cut that beat from the format.
• A flat curve with a spike at the end: your payoff works, market it earlier in the video.

Thirty minutes, once a week, same day:

1. 1Rank the week's videos by retention, not views
2. 2For the top video: name what was different: topic, hook style, pacing, visual
3. 3For the worst video: find the drop-off point on the curve
4. 4Write one change to the format and apply it to next week's batch
5. 5Check the idea backlog and refill it with variations of the winners

One deliberate format improvement per week is 12 improvements a quarter. That compounding is the whole game.

When a video breaks out, the audience voted. Make the sequel, make three more on adjacent topics, turn it into a recurring series inside the channel. Creators feel repetitive doing this; audiences experience it as the channel finding its identity. Your breakout format should dominate your schedule until the data says it is exhausted.