Message Engine

Performance Learning

Cross-platform video insights from 377 scored latest results. Blueprint v15 generated 27 Jul 26, 22:41.

Controlled evidence loop

Does Blueprint-guided generation outperform control?

There is not yet enough mature, matched, human-reviewed evidence to compare Blueprint and control ideas.

insufficient data

Blueprint ideas

Control ideas

Matched pairs

0/12

Mean effect

—

Calibration MAE

—

Blocking evidence deficits

matched pairs0/12
A paired comparison requires twelve distinct Blueprint/control pairs.
blueprint human rating coverage0/0.7
At least 70% of mature Blueprint ideas need blinded human ratings.
control human rating coverage0/0.7
At least 70% of mature control ideas need blinded human ratings.
blueprint reviewed candidates0/12
At least twelve Blueprint candidates must pass through the blinded selection ledger.
control reviewed candidates0/12
At least twelve control candidates must pass through the blinded selection ledger.

Highest-value next experiments

Publish 12 additional mature control ideas12 ideas
Twelve ideas in each cohort are needed to form twelve matched pairs; 0 mature control ideas currently qualify.
Publish 12 additional mature blueprint ideas12 ideas
Twelve ideas in each cohort are needed to form twelve matched pairs; 0 mature blueprint ideas currently qualify.
Run balanced blinded candidate reviews24 ideas
The candidate ledger needs adequate and comparably selected cohorts before published results can support a causal claim.
Complete blinded human quality ratings1.4 ideas
Human ratings prevent reach alone from defining idea quality.
Reduce uncertainty for redemptive reversal5 ideas
0/5 distinct mature ideas currently support this hypothesis.

Descriptive diagnostics — not a causal conclusion

Maximize the long-term quality of marketing ideas, not content volume.

A strong idea earns sustained attention, meaningful audience response, and human confidence while improving prediction accuracy over time.

Use the controlled evidence loop above for any Blueprint-versus-control claim.

Uncontrolled trend

Collecting evidence

Mature ideas

Human rated

Generation	Ideas	Quality	Win rate	Calibration error	Lift vs legacy
Before Blueprint	22	38.7	9%	32.3	—
Blueprint v1	35	45.0	20%	27.4	+6.3
Blueprint v14	0	—	—	—	—

Scored results

377

Pending

Make more

Test more

Avoid default

YouTube Blueprint

Platform-specific scorecard from the YouTube posts that Gemini ingested.

775 avg views

Based on 87 analysed performance results, optimize for the properties with the strongest repeatable lift, not isolated one-off winners.

Use in the next video

Make the hook Vague.
Anchor the topic in a strong personal-growth, AI, or business claim.
Use a room with moderate visual context rather than a flat wall.
Keep clothing simple and high contrast.
Use clear captions with emphasis where it helps the hook land.

Avoid as the default

Avoid defaulting to Grey clothing.
Avoid the Window Or Skylight setup as the default.

Suggested brief

Make the hook Vague. Anchor the topic in a strong personal-growth, AI, or business claim. Use a room with moderate visual context rather than a flat wall. Open with the title/caption visible in the first second. Keep the delivery direct and visually readable.

Overall Video Blueprint

Same video metadata joined to all platform performance.

344 platform results

The YouTube upload is used as the video file for Gemini, but the resulting visual and packaging labels describe the source video. This overall blueprint compares those labels against normalized performance across YouTube, TikTok, Instagram, and Facebook.

Do when posting everywhere

Title style: Text Only
+17%
Include Text Only when it fits; this property is running above its baseline.
Avg score 35 · 20 platform results
Posted hour: 18
+16%
Include 18 when it fits; this property is running above its baseline.
Avg score 35 · 47 platform results
Hook specificity: Specific
+9%
Include Specific when it fits; this property is running above its baseline.
Avg score 33 · 24 platform results
Target viewer: Creator
+8%
Include Creator when it fits; this property is running above its baseline.
Avg score 32 · 20 platform results
Setting: Bedroom
+8%
Include Bedroom when it fits; this property is running above its baseline.
Avg score 32 · 48 platform results
Headline format: Warning
+6%
Include Warning when it fits; this property is running above its baseline.
Avg score 32 · 34 platform results
Top type: Zip Up Fleece
+5%
Include Zip Up Fleece when it fits; this property is running above its baseline.
Avg score 31 · 48 platform results

Avoid as the cross-platform default

Posted hour: 8-14%
Title position: Chest Center-10%
Target viewer: Founder-9%
Expression: Serious-9%
Topic: Business Strategy-8%
Setting: Window Or Skylight-7%
Title word count: Seven Plus Words-6%
Contrast: High-5%

Overall suggested brief

Make the hook Specific. Anchor the topic in a strong personal-growth, AI, or business claim. Use a Bedroom setting. Open with the title/caption visible in the first second. Keep the delivery direct and visually readable.

Do / Don’t Signals

Each property is compared with the YouTube baseline using average views lift.

Do / include

Property	Lift	Avg views	Top Q	N
Hook specificity: Specific Include Specific when it fits; this property is running above its baseline.	+28%	996	33%	6
Title style: Text Only Include Text Only when it fits; this property is running above its baseline.	+23%	955	40%	5
Title case: All Caps Include All Caps when it fits; this property is running above its baseline.	+15%	892	29%	41
Target viewer: Creator Include Creator when it fits; this property is running above its baseline.	+14%	885	0%	5
Hook specificity: Vague Include Vague when it fits; this property is running above its baseline.	+14%	882	38%	13
Title position: Above Head Include Above Head when it fits; this property is running above its baseline.	+14%	880	29%	42
Title zone: Upper Third Include Upper Third when it fits; this property is running above its baseline.	+14%	880	29%	42
Headline format: Warning Include Warning when it fits; this property is running above its baseline.	+13%	875	33%	9

Don’t / exclude

Property	Lift	Avg views	Bottom Q	N
Top color: Grey Do not use Grey as the default; this property is running below its baseline.	-51%	383	71%	7
Title position: Chest Center Do not use Chest Center as the default; this property is running below its baseline.	-28%	561	50%	14
Topic: Business Strategy Do not use Business Strategy as the default; this property is running below its baseline.	-27%	570	40%	10
Title word count: Seven Plus Words Do not use Seven Plus Words as the default; this property is running below its baseline.	-26%	571	40%	10
Target viewer: Founder Do not use Founder as the default; this property is running below its baseline.	-24%	589	45%	11
Setting: Window Or Skylight Do not use Window Or Skylight as the default; this property is running below its baseline.	-20%	623	42%	19
Title case: Mixed Case Do not use Mixed Case as the default; this property is running below its baseline.	-16%	653	41%	37
Title zone: Middle Third Do not use Middle Third as the default; this property is running below its baseline.	-15%	659	39%	36

Methodology

How the won / redo / cold call is made.

Each idea is assumed to run everywhere. Results are normalized per platform first, then judged against baseline: green beat it, amber was close, blue missed it.

Decision threshold

Won somewhere55+

Sharpen and retest45-54

Hold or rewrite<45

Composite score

retention35%

engagement35%

reach30%

Engagement value

likes1x

replies3x

comments3x

reposts4x

shares4x

saves4x

Generation mix

Use strongest patterns70%

Test promising variants20%

Try new bets10%

Learning platforms: facebook, instagram, tiktok, YouTube Shorts. Results become eligible after 0 hours so early reach does not masquerade as durable quality.

Performance

Theme families ranked by latest cross-platform performance.

Theme family	Platforms	Results	Decision	Best	Win rate
visibility distribution attention alignment_over_scale, apprenticeship_to_jesus, childhood_clues_to_purpose, connection_and_flourishing	4	96	Redo · 46	74	30%
obedience and purpose circle_shapes_ceiling, coasting_mediocrity, gifts_multiply_through_use, next_step_faith	4	52	Cold · 44	76	27%
redemptive reversal redemptive_reversal	4	26	Redo · 48	77	38%
formation through discomfort analysis_paralysis_fear, external_validation, false_humility_fear, finishing_last_twenty_percent	4	52	Cold · 43	70	19%
builder product leverage alignment_over_usefulness, become_then_build, control_versus_surrender, creative_collision	4	51	Cold · 39	64	18%
ai capability and limits agentic_evaluation, agentic_leverage, ai_accountability, ai_creative_execution	4	32	Cold · 42	64	22%
constraint to assignment anchored_restriction, momentum_preservation, movement_creates_clarity	4	44	Cold · 40	77	16%
energy infrastructure systems ai_energy_constraint, environment_multiplying_or_draining, purpose_beyond_applause	4	16	Cold · 42	66	25%
control surrender peace control_vs_peace	4	8	Redo · 49	71	38%

Lift

-2.3