Sample data. This is a fictional demonstration workspace. Next: how findings become experiments →

Insights

Method mv1, window 2026-04-02 to 2026-07-01 , 20 posts considered , run 2026-07-01.

Calculated Result low confidence

In these 4 instagram video posts with duration under 15s, the median average view duration ratio was 55% higher than the 4 posts with duration 15 to 30s, between 2026-04-02 and 2026-07-01.

8 posts, 2026-04-02 to 2026-07-01. Medians 0.8479 vs 0.5475. No counterexamples in this sample.

How was this calculated?

This comparison is low confidence because the smaller group has 4 posts; the median difference is 55%; it survives removing the strongest post; the posts span more than 30 days. Higher confidence labels are disabled in method version 1 until its rules and fixtures are approved.

Observational comparison within this creator's own posts; it does not establish why the difference exists.

Formula: median average view duration ratio of the 4 posts with under 15s versus the 4 with 15 to 30s.

Consider testing this as a controlled experiment before treating it as a rule.

Inspect evidence and give your verdict

Calculated Result low confidence

In these 4 instagram still posts with a question opening, the median save rate by reach was 51% higher than the 8 posts with a statement opening, between 2026-04-02 and 2026-07-01.

12 posts, 2026-04-02 to 2026-07-01. Medians 0.0305 vs 0.0202. No counterexamples in this sample.

How was this calculated?

This comparison is low confidence because the smaller group has 4 posts; the median difference is 51%; it survives removing the strongest post; the posts span more than 30 days. Higher confidence labels are disabled in method version 1 until its rules and fixtures are approved.

Observational comparison within this creator's own posts; it does not establish why the difference exists.

Formula: median save rate by reach of the 4 posts with question versus the 8 with statement.

Consider testing this as a controlled experiment before treating it as a rule.

Inspect evidence and give your verdict

Calculated Result low confidence

In these 4 instagram video posts with a call to action, the median comment rate by reach was 50% higher than the 4 posts with no call to action, between 2026-04-02 and 2026-07-01.

8 posts, 2026-04-02 to 2026-07-01. Medians 0.0120 vs 0.0080. No counterexamples in this sample.

How was this calculated?

This comparison is low confidence because the smaller group has 4 posts; the median difference is 50%; it survives removing the strongest post; the posts span more than 30 days. Higher confidence labels are disabled in method version 1 until its rules and fixtures are approved.

Observational comparison within this creator's own posts; it does not establish why the difference exists.

Formula: median comment rate by reach of the 4 posts with true versus the 4 with false.

Consider testing this as a controlled experiment before treating it as a rule.

Inspect evidence and give your verdict

What would unlock more

  • average view duration ratio is unavailable for 12 of 12 posts in this group, so this comparison was not run.
  • follows reported per reach is unavailable for 12 of 12 posts in this group, so this comparison was not run.
  • Not enough comparable posts to evaluate topic for save rate by reach on instagram. Add at least 2 more video posts with topic set to gear to test this.
  • Not enough comparable posts to evaluate topic for comment rate by reach on instagram. Add at least 2 more video posts with topic set to gear to test this.
  • Not enough comparable posts to evaluate topic for share rate by reach on instagram. Add at least 2 more video posts with topic set to gear to test this.