# MY TIKTOK LEARNING LOG
Version: 1.0
Purpose: Evolving evidence record for one creator's TikTok account.

> This file becomes more important over time.
> Do not silently rewrite history.
> Record what was tested, what happened and how confident the conclusion is.

---

# 1. LEARNING STATES

Use these exact states:

## HYPOTHESIS
An idea worth testing.

## EMERGING PATTERN
A repeated signal, but not enough evidence to guide everything yet.

## ACCOUNT RULE
A pattern repeated across multiple comparable posts and useful enough to influence planning.

## RETIRED RULE
A previous belief no longer supported by live evidence.

One post is an anecdote.

Repeated comparable outcomes create learning.

---

# 2. STARTING RULES FROM THE MASTER SYSTEM

These are evidence-backed starting principles, not personal account rules yet.

## Viewer before company
Start with a human problem, desire, objection or situation before asking the viewer to care about the company.

## Specificity
Specific details are normally more useful than generic "freedom/success/opportunity" language.

## Curiosity with genuine payoff
Create a reason to stay, then honestly close the loop.

## Creator as proof
Real experience, process, mistakes and believable evidence are strategically stronger than constant hype.

## Native production is enough
A normal phone and ordinary environment can work.

## Mixed content jobs
Not every post should be a direct recruitment post.

## Personal evidence eventually wins
As this account's repeated data grows, it should influence decisions more heavily than generic advice.

---

# 3. MASTER-SYSTEM HYPOTHESES TO TEST

These came from the source research and should be tested rather than assumed.

## H1 — 61–90 second substantial content
**Status:** Hypothesis  
**Source:** UW cohort

The mature UW dataset showed the strongest median relative performance in the 61–90 second duration bucket.

Test this for content that genuinely earns the duration.

Do not stretch short ideas.

---

## H2 — Hard CTA vs reach
**Status:** Hypothesis  
**Source:** UW cohort

Direct CTA language was associated with lower median relative reach.

Test soft/no-CTA reach content against direct conversion content.

Do not interpret this as "CTAs are bad".

---

## H3 — Objection-led content
**Status:** Hypothesis  
**Source:** UW cohort

MLM/scam/pyramid objection wording showed an exploratory positive signal and was a major strength for one studied creator.

Test whether this fits this creator's audience and personality.

---

## H4 — Low-barrier specificity
**Status:** Hypothesis  
**Source:** UW cohort

Low-barrier wording showed an exploratory positive caption-theme signal.

Do not turn this into misleading earnings or effort claims.

---

# 4. CREATOR-SPECIFIC ACTIVE HYPOTHESES

Add personalised tests here.

## H5
**Hypothesis:**  
**Why we think this:**  
**Primary metric:**  
**Posts needed:**  
**Status:**  

## H6
**Hypothesis:**  
**Why we think this:**  
**Primary metric:**  
**Posts needed:**  
**Status:**  

## H7
**Hypothesis:**  
**Why we think this:**  
**Primary metric:**  
**Posts needed:**  
**Status:**  

---

# 5. POST EXPERIMENT LOG

Duplicate this section for every meaningful reviewed post.

## EXPERIMENT [ID]

**Date posted:**  
**Date reviewed:**  
**Post/title:**  
**Post link:**  

**Primary content job:**  
[Reach / Resonance / Authority / Trust / Community / Conversion]

**Audience:**  
**Content pillar:**  
**Format:**  
**Hook family:**  
**Exact hook:**  
**Video length:**  
**CTA:**  
**Posting time:**  
**Search phrase/topic:**  

### Hypothesis
What were we testing?

> 

### Variable(s) being tested
Change only one or two meaningful variables where possible.

> 

### Comparable posts / baseline

> 

### Mature analytics

**Views:**  
**Relative views vs baseline:**  
**Average watch time:**  
**Completion/full-watch rate:**  
**Retention notes:**  
**Likes:**  
**Comments:**  
**Shares:**  
**Saves/favourites:**  
**Profile visits:**  
**New followers:**  
**Traffic source:**  
**Search traffic/queries:**  

### Recruitment/commercial outcome

**Relevant DMs:**  
**Qualified conversations:**  
**Explainer sends:**  
**Information-call registrations:**  
**Calls held:**  
**Partner registrations:**  

### Funnel diagnosis
Choose primary area:

- [ ] STOP
- [ ] CARE
- [ ] STAY
- [ ] REWARD
- [ ] SHARE
- [ ] ACT

### What worked

> 

### What weakened it

> 

### Interpretation

> 

### Confidence

- [ ] Low
- [ ] Emerging
- [ ] Strong

### Decision

- [ ] Repeat
- [ ] Variation
- [ ] Test again
- [ ] Reduce
- [ ] Hold
- [ ] Retire
- [ ] Series candidate

### Exact next test

> 

### Learning update

> 

---

# 6. WEEKLY REVIEW

**Week commencing:**  

## Best-performing post
**Post:**  
**Why it appears to have worked:**  

## Most useful post for recruitment
**Post:**  
**Outcome:**  

## Weakest post
**Post:**  
**Likely reason:**  

## Strongest hook
> 

## Strongest topic/pillar
> 

## Strongest format
> 

## Strongest length
> 

## Best CTA/conversion route
> 

## Useful comments/questions to turn into content
1.  
2.  
3.  

## What to repeat next week
1.  
2.  
3.  

## What to test next week
1.  
2.  

---

# 7. MONTHLY REVIEW

**Month:**  

## Account baseline
**Number of posts:**  
**Median mature-post views:**  
**Median average watch time:**  
**Median completion rate:**  
**Median shares:**  
**Median saves:**  
**Median profile visits:**  
**Relevant follower growth:**  
**Qualified conversations:**  
**Information calls:**  
**Partner registrations:**  

## Best content pillars
1.  
2.  
3.  

## Best hook families
1.  
2.  
3.  

## Best formats
1.  
2.  
3.  

## Best-performing length bands
1.  
2.  

## Best audience/persona
> 

## Best conversion route
> 

## What repeatedly failed
> 

## What deserves more testing
> 

## Changes for next month
1.  
2.  
3.  

---

# 8. EMERGING PATTERNS

Only add something here after it has repeated across comparable posts.

## Pattern [P1]
**First observed:**  
**Repeated across:**  
**Evidence:**  
**What it appears to mean:**  
**Confidence:**  
**Next test:**  

---

# 9. ACCOUNT RULES

These are the most important learnings in the file.

Only promote a pattern when evidence is strong enough to influence future planning.

## Rule [R1]
**Rule:**  
**Evidence:**  
**Date promoted:**  
**Applies to:**  
**Exceptions/limits:**  

---

# 10. RETIRED RULES

Never silently delete old beliefs.

## Retired rule [RR1]
**Old rule:**  
**Why it was believed:**  
**Evidence that challenged it:**  
**Replacement rule:**  
**Date retired:**  

---

# 11. SERIES PERFORMANCE

## Series 1
**Name:**  
**Episodes published:**  
**Median views:**  
**Follower impact:**  
**Lead impact:**  
**Retention pattern:**  
**Decision:** Continue / refine / retire

## Series 2
**Name:**  
**Episodes published:**  
**Median views:**  
**Follower impact:**  
**Lead impact:**  
**Retention pattern:**  
**Decision:** Continue / refine / retire

---

# 12. POSTING-TIME LEARNING

Do not create a posting-time rule from one good post.

## Tested windows

| Day/window | Comparable posts | Relative performance | Lead/follower quality | Confidence |
|---|---:|---:|---|---|
|  |  |  |  |  |

## Current posting-time conclusion
> No established rule yet.

---

# 13. AUDIENCE LANGUAGE BANK

Capture real wording used in comments and DMs.

These phrases can be powerful future hooks because they come from the audience.

1.  
2.  
3.  
4.  
5.  

---

# 14. CONTENT IDEAS CREATED BY REAL AUDIENCE FEEDBACK

| Question / objection / comment | Content idea | Status |
|---|---|---|
|  |  |  |

---

# 15. VERSION HISTORY

## v1.0
Initial learning-log template created.

Future versions should record:
- major patterns promoted
- account rules changed
- retired assumptions
- material baseline changes