Platform analytics dashboards are designed to be looked at, not acted on — a wall of vanity metrics around four numbers that actually diagnose your clips. Here's the working clipper's reading of them, what each one tells you to fix, and the weekly ritual that turns data into hit rate.
1. First-second retention (the swipe test)
What share of viewers survive the opening moment. This is your hook grade, full stop.
- Low: the clip never got a chance — recut the first second, try a different pattern on the same moment, check that your text frame reads in one glance.
- High but views low: distribution problem, not hook problem — usually account profiling (niche consistency) or a saturated moment everyone else also cut.
2. Completion rate (the architecture grade)
Watched-to-end percentage — the algorithm's favorite signal and your edit grade.
- Falls off mid-clip: you left dead air or a tangent in; tighten.
- Falls off right before the end: you ended late — end AT the payoff.
- Over 100% / rewatch spikes: your loop works; do that seam again.
Watch the retention curve, not just the number: every dip timestamps an editing mistake for you.
3. Shares (the social-function grade)
Did the clip give viewers something to do — send it, quote it, tag someone? Shares punch far above their count because platforms weight them heavily (Reels most explicitly). Low shares on high completion means the clip entertained but didn't say anything — fine on TikTok, a miss on Reels and X.
4. Follows-per-view (the account grade)
New follows divided by reach tells you whether the clip built the account or just spent a slot machine pull. Formats that convert follows are your repeatable format candidates — they compound; pure-spike formats don't.
What to ignore
Total view counts in isolation (distribution luck reads identically to skill in a single number), likes (weakest signal on every platform), profile visits without follows, and any metric averaged across mixed niches or formats — averages hide the pattern you're hunting.
The weekly ritual (30 minutes, non-negotiable)
From your clip log: sort the week's clips by each of the four numbers. For the top and bottom two in each: name the cause in one sentence (hook pattern, moment type, edit choice, platform). Three weeks of this and your private playbook — this hook on this moment type for this platform — writes itself. That playbook is the entire difference between the hit rates in the earnings math.
Campaign context makes the numbers monetary: retention and completion decide which clips cross the view gate, and the gate decides which clips get paid on every view. Analytics isn't homework — it's the audit trail of your own money.