Kill Criteria: When to Retire an Underperforming Teach Reel
Retire, refresh, or merge weak teach Reels using signals beyond views—completion cliffs, saves, comment quality—and compare against the approved board intent.
Kill Criteria: When to Retire an Underperforming Teach Reel
Decide when to stop boosting or regenerating a weak teach asset. Views alone lie; completion cliffs, saves, and comment quality tell you whether the board intent failed.
Growth marketers and personal-IP creators reviewing weekly Reel performance need a retire/refresh/merge policy—not vibes. Measurement beyond views is the foundation (how to measure explainer Reels beyond views); this piece turns those signals into kill criteria.
Signals beyond raw views
Track a small set consistently (use your platform’s real metrics—do not invent benchmarks):
| Signal | What it suggests | Kill-relevant? |
|---|---|---|
| 3s hold / thumb-stop | Hook/frame fail | Refresh hook first |
| Completion / quartile drop | Mid-teach friction | Refresh pacing or split intent |
| Saves / shares | Utility | Low → weak teach value |
| Profile visits / follows | Curiosity about creator | Context only |
| Comment quality | Confusion vs application | Confusion → board rewrite |
| Boost CPA / cost per completed view | Paid inefficiency | Stop boost; decide organic fate |
A Reel can look “viral” on views and still fail teach intent if saves are near zero and comments ask basic clarifications you thought you covered.
Completion cliff patterns
- Cliff at 0–3s: first-frame / hook problem (first-frame clarity; non-clickbaity hooks)
- Cliff mid-method: too many steps, motion theater, or caption overload
- Cliff at disclaimer: disclosure weight or placement issue
- Smooth curve, low saves: entertaining but not useful—intent mismatch
Refresh vs retire vs merge
Refresh when the intent is still right but packaging failed:
- New hook on a locked body (A/B hooks on locked body)
- Still vs motion retags on muddy beats
- Caption rewrite for mute-first clarity
- UI skin update after product change (update storyboards mid-campaign)
Retire when:
- Intent is obsolete (feature gone, policy changed)
- Two refreshes failed the same signals
- Comments reveal the audience does not want this teach
- Legal pulls the claim set
- The Reel cannibalizes a stronger URL/intent without adding a new angle
Merge intents when:
- Two Reels teach overlapping steps and split attention
- A FAQ cluster works better as one clearer micro-intent
- Personal-IP calendar shows duplicate micro-intents (micro-intent calendar)
Merging means a new board with one promise—not stitching two weak edits together.
Compare against the approved board intent
Before regenerating blindly, reopen the approved storyboard and ask:
- Did viewers receive the intent we boarded?
- Which beat aligns with the completion cliff?
- Is the failure creative (hook/visual) or strategic (wrong intent)?
- Would a library module swap fix it, or do we need a new ID?
Regenerating the same board with a new aesthetic rarely fixes a wrong intent. Prompt-only retries without board diagnosis create prompt debt.
Weekly review ritual (30–45 minutes)
- Sort last week’s teaches by completed views and saves
- Flag bottom quartile for diagnosis—not automatic delete
- Label each: refresh / retire / merge / hold
- Log the board ID and the failing signal
- Stop paid boost on retire candidates immediately
Worked examples (pattern-level)
Example A — hook failure High impressions, poor 3s hold, decent saves among completers. Action: refresh hook variants on locked body; do not retire intent yet.
Example B — wrong intent Solid hold, steep mid cliff, comments ask a different question than the board promised. Action: merge/retire; spawn a new micro-intent board that matches the comments.
Example C — obsolete UI Performance was fine until product shipped; now comments say “menu moved.” Action: refresh UI skins on the same intent; update the board (mid-campaign UI updates).
Example D — paid fatigue Organic saves OK; boost CPA worsened for two weeks. Action: stop boost; keep organic; do not confuse paid fatigue with teach failure.
Library hygiene after kills
When you retire a Reel, update the beat library:
- Deprecate modules blamed for confusion
- Keep high-performing modules with last-used dates
- Note “do not combine with” if a metaphor failed in context
Stakeholder reviews should see kill notes beside boards (review rituals) so leadership does not demand endless regenerations of a dead intent.
Do not confuse seasonality with death
A teach that underperforms in a holiday week may rebound. Require at least two comparable windows before retire—unless comments prove intent mismatch immediately. Hold labels exist for a reason.
Soft production tip
Compare performance against the approved board intent—do not regenerate blindly. Storyboard-first practice (Scro included) keeps the intent artifact available so analytics conversations stay tied to beats, not only to finished MP4s.
Kill-criteria checklist
- Metrics defined beyond views
- Cliff mapped to a beat
- Refresh attempted with a hypothesis
- Retire/merge rules written down
- Boost stopped when criteria hit
- Library modules updated or deprecated
Retiring weak teaches is how strong libraries stay strong. Sentimentality is not a distribution strategy.