Marketing Analyst

def calculate_attribution(touchpoints, model='position'):
    """Calculate attribution credit for a conversion journey.

    Args:
        touchpoints: List of channel names in order of interaction
        model: One of 'first', 'last', 'linear', 'time_decay', 'position'

    Returns:
        Dict mapping channel -> credit (sums to 1.0)

    Example:
        >>> calculate_attribution(['paid_search', 'email', 'organic', 'direct'], 'position')
        {'paid_search': 0.4, 'email': 0.1, 'organic': 0.1, 'direct': 0.4}
    """
    n = len(touchpoints)
    credits = {}

    if model == 'first':
        credits[touchpoints[0]] = 1.0
    elif model == 'last':
        credits[touchpoints[-1]] = 1.0
    elif model == 'linear':
        for tp in touchpoints:
            credits[tp] = credits.get(tp, 0) + 1.0 / n
    elif model == 'time_decay':
        decay = 0.7
        total = sum(decay ** i for i in range(n))
        for i, tp in enumerate(reversed(touchpoints)):
            credits[tp] = credits.get(tp, 0) + (decay ** i) / total
    elif model == 'position':
        if n == 1:
            credits[touchpoints[0]] = 1.0
        elif n == 2:
            credits[touchpoints[0]] = 0.5
            credits[touchpoints[-1]] = credits.get(touchpoints[-1], 0) + 0.5
        else:
            credits[touchpoints[0]] = 0.4
            credits[touchpoints[-1]] = credits.get(touchpoints[-1], 0) + 0.4
            for tp in touchpoints[1:-1]:
                credits[tp] = credits.get(tp, 0) + 0.2 / (n - 2)

    return credits

# Campaign Analysis: Q1 2026 Product Launch

## Performance Summary
| Metric       | Target  | Actual  | vs Target |
|--------------|---------|---------|-----------|
| Impressions  | 500K    | 612K    | +22%      |
| Clicks       | 25K     | 28.4K   | +14%      |
| Leads        | 1,200   | 1,350   | +13%      |
| MQLs         | 360     | 410     | +14%      |
| Pipeline     | $1.2M   | $1.45M  | +21%      |
| Revenue      | $380K   | $425K   | +12%      |

## Channel Breakdown
| Channel      | Spend   | Leads | CPL   | Pipeline |
|--------------|---------|-------|-------|----------|
| Paid Search  | $45K    | 520   | $87   | $580K    |
| LinkedIn Ads | $30K    | 310   | $97   | $420K    |
| Email        | $5K     | 380   | $13   | $350K    |
| Content/SEO  | $8K     | 140   | $57   | $100K    |

## Key Insight
Email delivers lowest CPL ($13) and strong pipeline. Recommend shifting
10% of LinkedIn budget to email nurture sequences for Q2.

Budget Allocation Recommendation
  Channel        Current    Optimal    Change    Expected ROI
  Paid Search    30%        35%        +5%       4.2x
  Social Paid    25%        20%        -5%       2.8x
  Display        15%        10%        -5%       1.5x
  Email          10%        15%        +5%       8.5x
  Content        10%        12%        +2%       5.2x
  Events         10%        8%         -2%       2.2x

  Projected Impact: +15% pipeline with same budget

from scipy import stats
import numpy as np

def analyze_ab_test(control_conv, control_total, treatment_conv, treatment_total, alpha=0.05):
    """Analyze A/B test for statistical significance.

    Example:
        >>> result = analyze_ab_test(150, 5000, 195, 5000)
        >>> result['significant']
        True
        >>> f"{result['lift_pct']:.1f}%"
        '30.0%'
    """
    p_c = control_conv / control_total
    p_t = treatment_conv / treatment_total
    p_pool = (control_conv + treatment_conv) / (control_total + treatment_total)
    se = np.sqrt(p_pool * (1 - p_pool) * (1/control_total + 1/treatment_total))
    z = (p_t - p_c) / se
    p_value = 2 * (1 - stats.norm.cdf(abs(z)))

    return {
        'control_rate': p_c,
        'treatment_rate': p_t,
        'lift_pct': ((p_t - p_c) / p_c) * 100,
        'p_value': p_value,
        'significant': p_value < alpha,
    }