Short video produces confident-looking numbers that mean less than they appear to. Careful reels performance tracking compares like with like and resists drawing conclusions from any single video.
Completion over views
Views measure distribution, which depends heavily on factors outside your control. Completion and rewatches measure whether the video worked for the people it reached, which is the part your decisions affect.
A video with high views and low completion is usually a packaging success and a content failure. Treating it as an outright success leads to making more of the same, and the pattern compounds until the account is reaching many people who never stay.
What to record
- One row per video: date, format, theme, hook type.
- Views alongside completion, never views alone.
- Saves and shares, which indicate durable value.
- Follower change attributable to the period, not to a video.
- Flag outliers explicitly in both directions.
- Group by format monthly and compare group medians.
- Note anything unusual — a repost, a collaboration, paid support.
Group comparison is the point of the log. Ten videos in a format tell you something; one video tells you almost nothing, however dramatic its numbers.
Handle outliers explicitly
One video reaching an unusual audience can dominate a month and make everything else look like failure. Name it in the report and show the period without it, so nobody concludes that a general change occurred when a single video moved.
Be equally careful with a bad month. Short-form variance is wide enough that a poor month can occur with no change in quality at all, and rebuilding a working approach on that basis is a common and expensive mistake. Reporting structure is in the analytics reporting guide, and planning in the reels content calendar workflow.
Common questions
How long before results settle?
Short video can be surfaced for weeks. Allow time before judging, since early figures often understate the total.
Is follower growth a good measure?
Loosely, and only over months. Much reach comes from non-followers, so the relationship is weaker than it seems.
Should every video be reviewed?
Review outliers and sample the rest. Mid-range videos rarely have anything individual to teach.
What about comparing platforms?
Compare each against its own history. Cross-platform comparisons mostly measure how the platforms differ.
How often should the log be read?
Weekly to record, monthly to interpret. Reading it daily produces reactions to noise. See about.
Should the log include videos that were deleted?
Record them and why. A deleted video is still a data point, and removing the row makes the month look better than it was.
Should paid-supported videos sit in the same log?
Log them, flagged. Mixing promoted and organic results unflagged inflates the group medians and quietly misleads the next planning round.