Menü
Score 3 content variations and rank them with comparative analysis
Deep analysis of a specific segment including roles, topics, and past performance
Display high-level summary of audience segments, sizes, topics, and activity
Review calibration metrics, accuracy trends, and model confidence levels
Core skill for scoring a draft post with per-segment predictions and reasoning
| name | compare-variations |
| description | Score 3 content variations and rank them with comparative analysis |
| version | 1.0.0 |
Purpose: Score multiple post variations and compare them to identify the best performer.
{
"variation_1": "string (draft text)",
"variation_2": "string (draft text)",
"variation_3": "string (draft text)"
}
{
"ranked_variations": [
{
"rank": 1,
"overall_score": 85,
"confidence": "medium",
"predicted_reactions": 38,
"hook_type": "story",
"topics": ["Leadership", "Founder Stories"],
"top_segments": [
{
"segment": "Executives",
"predicted_reactions": 28,
"hook_strength": 92
}
],
"recommendation": "This is your best variation. Story hook + founder journey resonates strongly with Executives and Growth & Sales."
}
],
"comparative_analysis": {
"variation_1_vs_2": "Variation 1 scores 10 points higher. The story hook (92/100) beats the question hook (68/100) for your Executives.",
"variation_2_vs_3": "Variation 2 is slightly stronger. More technical depth appeals to your largest engaged segment (Technical Leaders).",
"winner": "Variation 1"
},
"hook_comparison": {
"story": 85,
"question": 75,
"data-driven": 68
},
"segment_coverage": {
"Executives": {
"best_variation": 1,
"reactions_sum": 28,
"hook_alignment": "story"
}
},
"recommendation": "Post Variation 1. The story hook and founder focus will drive engagement across Executives (your most engaged segment) and Growth & Sales. Expected performance: 38 reactions, strong resonance."
}
def compare_variations(variation_1, variation_2, variation_3):
"""Score and rank 3 variations."""
conn = get_db_connection()
try:
variations = [variation_1, variation_2, variation_3]
scores = []
# Score each variation independently
for i, draft in enumerate(variations, 1):
# Use score_draft skill to score this variation
score_result = score_draft(draft, track_for_calibration=False)
if 'error' in score_result:
return score_result
scores.append({
'variation_num': i,
'overall_score': score_result['overall_score'],
'confidence': score_result['confidence'],
'predicted_reactions': score_result['predicted_metrics']['reactions'],
'predicted_comments': score_result['predicted_metrics']['comments'],
'predicted_reach': score_result['predicted_metrics']['reach'],
'hook_type': score_result['hook_type'],
'topics': score_result['topics'],
'per_segment': score_result['per_segment']
})
# Sort by overall score (descending)
scores.sort(key=lambda s: s['overall_score'], reverse=True)
# Build ranked output
ranked_variations = []
for rank, score in enumerate(scores, 1):
top_segments = sorted(
score['per_segment'],
key=lambda p: p['predicted_reactions'],
reverse=True
)[:3]
recommendation = generate_variation_recommendation(
rank, score['overall_score'], score['hook_type'], top_segments
)
ranked_variations.append({
'rank': rank,
'variation_number': score['variation_num'],
'overall_score': score['overall_score'],
'confidence': score['confidence'],
'predicted_reactions': score['predicted_reactions'],
'predicted_comments': score['predicted_comments'],
'predicted_reach': score['predicted_reach'],
'hook_type': score['hook_type'],
'topics': score['topics'],
'top_segments': [
{
'segment': s['segment'],
'predicted_reactions': s['predicted_reactions'],
'hook_strength': s['hook_strength']
}
for s in top_segments
],
'recommendation': recommendation
})
# Comparative analysis
comparative = {
f"variation_{scores[0]['variation_num']}_vs_{scores[1]['variation_num']}":
f"Variation {scores[0]['variation_num']} scores {scores[0]['overall_score'] - scores[1]['overall_score']:.0f} points higher. "
f"Hook strength: {scores[0]['hook_type']} ({get_avg_hook_strength(scores[0]['per_segment']):.0f}) vs "
f"{scores[1]['hook_type']} ({get_avg_hook_strength(scores[1]['per_segment']):.0f}). "
f"Primary difference: {get_variation_difference(scores[0], scores[1])}.",
'winner': f"Variation {scores[0]['variation_num']}"
}
# Hook comparison
hook_comparison = {}
for score in scores:
hook = score['hook_type']
avg_strength = get_avg_hook_strength(score['per_segment'])
if hook not in hook_comparison:
hook_comparison[hook] = avg_strength
# Segment coverage
segment_coverage = {}
for seg_name in set(s['segment'] for score in scores for s in score['per_segment']):
best_var = max(scores, key=lambda s: next(
(p['predicted_reactions'] for p in s['per_segment'] if p['segment'] == seg_name), 0
))
segment_coverage[seg_name] = {
'best_variation': best_var['variation_num'],
'reactions_sum': sum(
p['predicted_reactions'] for p in best_var['per_segment']
if p['segment'] == seg_name
),
'hook_alignment': best_var['hook_type']
}
final_recommendation = (
f"Post Variation {scores[0]['variation_num']}. "
f"Score: {scores[0]['overall_score']}/100. "
f"The {scores[0]['hook_type']} hook and focus on "
f"{', '.join(scores[0]['topics'])} will drive {scores[0]['predicted_reactions']} reactions. "
f"Best resonance with {scores[0]['per_segment'][0]['segment']}."
)
return {
'ranked_variations': ranked_variations,
'comparative_analysis': comparative,
'hook_comparison': hook_comparison,
'segment_coverage': segment_coverage,
'recommendation': final_recommendation
}
except Exception as e:
return {'error': f'Error comparing variations: {str(e)}'}
finally:
conn.close()
def generate_variation_recommendation(rank, score, hook_type, top_segments):
"""Generate a human-readable recommendation for a variation."""
if rank == 1:
segments_str = ', '.join(s['segment'] for s in top_segments)
return (
f"This is your best variation. "
f"{hook_type.capitalize()} hook scores well ({score}/100). "
f"Strongest appeal: {segments_str}."
)
elif rank == 2:
return (
f"Solid alternative. Score: {score}/100. "
f"Consider this if you want to test a different hook ({hook_type}) or topic."
)
else:
return (
f"This variation underperforms the others ({score}/100). "
f"Consider revising the hook or topic angle, or choose variation 1-2."
)
def get_avg_hook_strength(per_segment):
"""Calculate average hook strength across segments."""
strengths = [p['hook_strength'] for p in per_segment]
return sum(strengths) / len(strengths) if strengths else 50
def get_variation_difference(var1, var2):
"""Describe the key difference between two variations."""
hooks_differ = var1['hook_type'] != var2['hook_type']
topics_differ = var1['topics'] != var2['topics']
if hooks_differ and topics_differ:
return f"different hook ({var1['hook_type']} vs {var2['hook_type']}) and different topics"
elif hooks_differ:
return f"different hook ({var1['hook_type']} vs {var2['hook_type']})"
elif topics_differ:
return f"different topics ({var1['topics']} vs {var2['topics']})"
else:
return "similar structure but different wording"
{
"ranked_variations": [
{
"rank": 1,
"variation_number": 1,
"overall_score": 85,
"confidence": "medium",
"predicted_reactions": 38,
"predicted_comments": 9,
"predicted_reach": 2660,
"hook_type": "story",
"topics": ["Leadership", "Founder Stories"],
"top_segments": [
{
"segment": "Executives",
"predicted_reactions": 28,
"hook_strength": 92
},
{
"segment": "Growth & Sales",
"predicted_reactions": 7,
"hook_strength": 80
}
],
"recommendation": "This is your best variation. Story hook scores well (85/100). Strongest appeal: Executives, Growth & Sales."
},
{
"rank": 2,
"variation_number": 2,
"overall_score": 75,
"confidence": "medium",
"predicted_reactions": 28,
"predicted_comments": 7,
"predicted_reach": 1960,
"hook_type": "question",
"topics": ["Leadership"],
"top_segments": [
{
"segment": "Technical Leaders",
"predicted_reactions": 12,
"hook_strength": 82
}
],
"recommendation": "Solid alternative. Score: 75/100. Consider if you want to test a different hook (question) or topic."
}
],
"comparative_analysis": {
"variation_1_vs_2": "Variation 1 scores 10 points higher. Hook strength: story (92/100) vs question (82/100). Primary difference: different hook (story vs question).",
"winner": "Variation 1"
},
"hook_comparison": {
"story": 85,
"question": 75,
"data-driven": 68
},
"segment_coverage": {
"Executives": {
"best_variation": 1,
"reactions_sum": 28,
"hook_alignment": "story"
}
},
"recommendation": "Post Variation 1. Score: 85/100. The story hook and focus on Leadership, Founder Stories will drive 38 reactions. Best resonance with Executives."
}