The iteration loop

Goal

Turn analytics into improvement by running small, deliberate experiments and recording what you learn.

Why most people don’t improve

People post, glance at the numbers, feel good or bad, and move on without changing anything on purpose. Improvement needs a loop: observe, hypothesize, change one thing, publish, compare, record. Doing this across 10-20 reels teaches you more than any course.

The loop

1. Observe

Use your analytics table (lesson 7.2). Identify one pattern: “My first-3-second retention is lower on videos that open with ‘So,’.”

2. Hypothesize

Write it as a testable statement: “If I open with the result instead of a greeting, hook retention will rise.”

3. Change one variable

Choose one: the hook, the first frame, the length, the pace, the caption style, the cover, the angle, the topic. Change only that. If you change the hook and the length together, you won’t know which mattered.

4. Publish

Publish the variant under similar conditions: similar topic, similar time of day, same platform.

5. Compare

After 24 hours and 7 days, compare the relevant metric to your recent baseline. Don’t compare to your best or worst; use the median of your last 5-10 reels.

6. Record

Write the result in your experiment log, even when the result is “no difference”. Negative results are valuable.

The experiment log

The numbers below are made up, for illustration only.

#	Hypothesis	Variable changed	Videos	Baseline	Result	Decision
1	Result-first hooks raise 3-s retention	Hook type	3	median 52%	61%	Keep
2	Shorter reels (under 25 s) raise completion	Length	3	median 38%	38%	No change

A decision column forces you to act: keep, drop or retest.

Ideas for experiments

• Hook type: question vs result vs bold claim.
• First frame: face vs object vs text.
• Length: 15-20 s vs 30-40 s.
• Pace: with and without jump cuts.
• Captions: highlight style, position.
• Opening text: with a headline vs without.
• Format: story vs list.
• Cover: face vs text.
• CTA: a question vs none.
• Comment keyword: “comment [word] and I’ll send you [resource]” vs no ask. Test where it sits too: in the hook, in the last 3-4 seconds on screen, or only in the caption.
• Content type: an expert reel vs a personal one on the same topic (lesson 2.2).
• Posting time: morning vs evening.
• Angle: eye-level vs low.

Pick one per week at first.

Sample size

One video proves little. For each experiment, aim for at least 3 videos per variant when you can. In the first weeks you may only have enough for rough signals; treat them as hints, not proof.

Keep the core

Experiments change the details. They do not change who you are. If the data says shorter, more upbeat videos perform better and that makes you feel like a fake, weigh the cost. Sustainable creators find the overlap between what the audience rewards and what they can do without hating it.

When something works

• Document the pattern in a “playbook” note: what you did, the numbers, why you think it worked.
• Repeat it in a new context, with a different topic. If it works again, it’s a pattern.
• Don’t over-copy it. Variation keeps the feed fresh.

When nothing works

• Check the basics: audio, light, hook, length.
• Check the thread: is it clear who the videos are for?
• Watch a few of your videos as a stranger. Where would you leave?
• Ask three people you trust to watch a reel and tell you where they got bored.
• Be patient. Four weeks is a short time; most skills show up gradually.

Weekly review (30 minutes)

Once a week:

1. Update the analytics table.
2. Look at the best and worst reel of the week.
3. Update the experiment log.
4. Choose next week’s single experiment.
5. Choose your three concepts for the week.
6. Celebrate the streak.

Common mistakes

• Changing many things.
• Testing without writing.
• Quitting an experiment too early.
• Letting data override your voice entirely.
• Forgetting to celebrate progress.

Exercise

1. Write your first hypothesis.
2. Set up the experiment log.
3. Plan the variants for your next three reels.
4. Schedule your weekly review.

Daily practice

• Minimum (keeps the streak): a voice note of any length.
• Today’s take (about 30 seconds): Film a variant of an earlier take with exactly one thing changed.
• Weeks 1-2 everything stays private; from week 3 follow your publishing plan.
• Don’t watch the take right away (lesson 4.1). Tick the day in your tracker.

Reading (optional)

This week’s book, about 15-20 minutes: Jonah Berger, Contagious. A loose suggestion; it does not have to match the lesson. More in resources and the schedule.

Checklist

• I have a hypothesis and one variable.
• I keep an experiment log with a decision column.
• I compare to a baseline median.
• I have a weekly review slot.
• I recorded today’s voice note and take.

---

← 7.2: Reading analytics | Course index | 8.1: Systems for consistency →