Short-form data swings violently, which makes it easy to tell almost any story with it. Disciplined tiktok analytics reporting fixes the metrics and the window in advance, so a month cannot be narrated by whichever number happened to look best.
Averages hide everything
A handful of videos usually account for most views, so the average tells you little about the typical video. Report the median alongside it, or show the distribution, and the picture becomes far more honest.
Watch time and completion matter more than views. A video seen by many people who left immediately is not a success, and treating it as one leads to making more of the same. Views measure distribution; completion measures whether the video worked.
The routine
- Fix the metric set and keep it even in poor months.
- Use identical window lengths so periods compare.
- Report median as well as average.
- Name outlier videos explicitly and show the period without them.
- Pair completion with views for every video discussed.
- Track follower growth against video count published.
- Close with decisions and owners, not observations.
Do not over-read a single video
Variance between comparable short videos is enormous. Two videos made the same way on the same subject can differ by an order of magnitude, so drawing conclusions from one is close to guessing.
Look at groups instead: all videos using a format, all videos on a theme. Patterns across ten videos are worth acting on, where a single result is usually noise wearing the costume of a signal. The stakeholder-facing version is covered in performance tracking, and the account-level review in the profile audit checklist.
Common questions
How often should reporting happen?
Weekly for the team, monthly for anything reported upward. Weekly stakeholder reporting mostly reports variance.
Is follower count a useful measure?
Loosely. Much reach comes from non-followers, so it describes less than it does on other platforms.
Should each video be analysed?
Analyse the outliers and sample the rest. Individual mid-range videos rarely have anything to teach.
What about comparing to other accounts?
Published benchmarks vary too widely to be useful. Your own history is the fair comparison.
How long before a video's numbers settle?
Short video can be surfaced for weeks, so allow time before judging. Early figures often understate the total. See contact.
What should be reported when a video does unusually well?
Say what was different about it, and be honest when nothing was. Attributing an outlier to a decision nobody actually made leads to repeating the wrong thing for months.
Should completion be compared across video lengths?
No. Shorter videos complete more often almost by definition, so compare within similar lengths or the ranking simply sorts by duration.