HubSpot AI营销自动化平台

import numpy as npfrom collections import defaultdict # HubSpot AI Marketing 营销归因模拟 # 模拟客户转化路径数据conversion_paths = [    {"touchpoints": ["SEO博客", "邮件营销", "产品演示"], "converted": True, "value": 5000},    {"touchpoints": ["社媒广告", "SEO博客", "白皮书下载", "销售跟进"], "converted": True, "value": 8000},    {"touchpoints": ["直接访问", "产品演示"], "converted": True, "value": 3000},    {"touchpoints": ["邮件营销", "社媒广告", "SEO博客"], "converted": False, "value": 0},    {"touchpoints": ["SEO博客", "邮件营销", "社媒广告", "产品演示", "销售跟进"], "converted": True, "value": 12000},    {"touchpoints": ["社媒广告", "直接访问"], "converted": False, "value": 0},    {"touchpoints": ["白皮书下载", "邮件营销", "产品演示"], "converted": True, "value": 6000},    {"touchpoints": ["SEO博客", "白皮书下载", "邮件营销"], "converted": False, "value": 0},] def lead_score(behavior_points, company_size, engagement_count):    """模拟HubSpot AI线索评分模型"""    behavior_score = min(behavior_points / 100, 1.0) * 40    size_score = min(company_size / 500, 1.0) * 30    engage_score = min(engagement_count / 10, 1.0) * 30    return round(behavior_score + size_score + engage_score, 1) def first_touch_attribution(paths):    """首次触点归因"""    attribution = defaultdict(float)    for p in paths:        if p["converted"] and p["touchpoints"]:            attribution[p["touchpoints"][0]] += p["value"]    return dict(attribution) def last_touch_attribution(paths):    """末次触点归因"""    attribution = defaultdict(float)    for p in paths:        if p["converted"] and p["touchpoints"]:            attribution[p["touchpoints"][-1]] += p["value"]    return dict(attribution) def linear_attribution(paths):    """线性归因"""    attribution = defaultdict(float)    for p in paths:        if p["converted"] and p["touchpoints"]:            share = p["value"] / len(p["touchpoints"])            for tp in p["touchpoints"]:                attribution[tp] += share    return dict(attribution) print("=== 多触点归因模型对比 ===")total_value = sum(p["value"] for p in conversion_paths if p["converted"])print(f"总转化价值: ${total_value}")print() models = {    "首次触点归因": first_touch_attribution(conversion_paths),    "末次触点归因": last_touch_attribution(conversion_paths),    "线性归因": linear_attribution(conversion_paths),} for model_name, result in models.items():    print(f"--- {model_name} ---")    for channel, value in sorted(result.items(), key=lambda x: -x[1]):        print(f"  {channel}: ${value:,.0f} ({value/total_value*100:.1f}%)")    print() print("=== 线索评分演示 ===")leads = [    {"name": "企业A", "behavior": 85, "size": 200, "engagement": 7},    {"name": "企业B", "behavior": 30, "size": 50, "engagement": 2},    {"name": "企业C", "behavior": 60, "size": 800, "engagement": 5},    {"name": "企业D", "behavior": 95, "size": 300, "engagement": 9},] for lead in leads:    score = lead_score(lead["behavior"], lead["size"], lead["engagement"])    grade = "A" if score >= 80 else "B" if score >= 60 else "C" if score >= 40 else "D"    print(f"{lead['name']}: 评分={score}, 等级={grade}")    print(f"  -> {'高价值线索，触发自动培育' if score >= 60 else '普通线索，持续观察'}")print() print("=== AB测试显著性模拟 ===")def ab_test_significance(va, ca, vb, cb):    pa, pb = ca/va, cb/vb    p_pool = (ca + cb) / (va + vb)    se = np.sqrt(p_pool * (1 - p_pool) * (1/va + 1/vb))    z = (pb - pa) / se    p_value = 2 * (1 - np.abs(z)) if abs(z) < 3 else 0.001    return pa, pb, p_value pa, pb, p_val = ab_test_significance(5000, 520, 5000, 610)print(f"版本A: 转化率={pa*100:.2f}%")print(f"版本B: 转化率={pb*100:.2f}%")print(f"提升率: {(pb-pa)/pa*100:.1f}%")print(f"统计显著: {'是 (p<0.05)' if p_val < 0.05 else '否'}")