如何通过 SERP API 跟踪 Google AI 模式响应

import os, requests, json
from datetime import datetime

API_KEY = os.environ['SCAVIO_API_KEY']
SH = {'x-api-key': API_KEY, 'Content-Type': 'application/json'}

BRAND = 'Scavio'
KEYWORDS = [
    'best search api for ai agents',
    'how to add search to ai agent',
    'mcp search tool',
    'serp api alternative',
    'web search api pricing',
]

def check_ai_mode(keyword, brand):
    data = requests.post('https://api.scavio.dev/api/v1/search',
        headers=SH, json={'query': keyword, 'country_code': 'us'}, timeout=10).json()
    # Check AI overview / featured snippet
    ai_overview = data.get('ai_overview', data.get('answer_box', {}))
    organic = data.get('organic_results', [])
    featured = data.get('featured_snippet', {})
    brand_lower = brand.lower()
    in_ai = brand_lower in json.dumps(ai_overview).lower() if ai_overview else False
    in_featured = brand_lower in json.dumps(featured).lower() if featured else False
    in_organic = any(brand_lower in json.dumps(r).lower() for r in organic[:10])
    organic_pos = next((i+1 for i, r in enumerate(organic) if brand_lower in json.dumps(r).lower()), None)
    return {
        'keyword': keyword,
        'in_ai_mode': in_ai,
        'in_featured': in_featured,
        'in_organic': in_organic,
        'organic_position': organic_pos,
        'has_ai_overview': bool(ai_overview),
    }

print(f'Tracking AI Mode for "{BRAND}" across {len(KEYWORDS)} keywords\n')
results = []
for kw in KEYWORDS:
    r = check_ai_mode(kw, BRAND)
    results.append(r)
    ai_status = 'CITED' if r['in_ai_mode'] else 'ABSENT'
    org_status = f'#{r["organic_position"]}' if r['organic_position'] else 'absent'
    print(f'  {kw[:40]:40} | AI: {ai_status:6} | Organic: {org_status}')
print(f'\nCost: ${len(KEYWORDS) * 0.005:.3f}')

def ai_mode_visibility(results, brand):
    total = len(results)
    ai_cited = sum(1 for r in results if r['in_ai_mode'])
    featured = sum(1 for r in results if r['in_featured'])
    organic = sum(1 for r in results if r['in_organic'])
    has_ai = sum(1 for r in results if r['has_ai_overview'])
    ai_score = (ai_cited / has_ai * 100) if has_ai else 0
    overall_score = ((ai_cited * 3 + featured * 2 + organic) / (total * 3) * 100)
    print(f'\n=== AI Mode Visibility: {brand} ===')
    print(f'  Keywords tracked:    {total}')
    print(f'  AI Mode present:     {has_ai}/{total} queries')
    print(f'  Brand in AI Mode:    {ai_cited}/{has_ai} ({ai_score:.0f}%)')
    print(f'  Brand in Featured:   {featured}/{total}')
    print(f'  Brand in Organic:    {organic}/{total}')
    print(f'  Overall Visibility:  {overall_score:.0f}/100')
    # Gaps
    gaps = [r['keyword'] for r in results if r['has_ai_overview'] and not r['in_ai_mode']]
    if gaps:
        print(f'\n  AI Mode Gaps (present but not cited):')
        for g in gaps:
            print(f'    - {g}')
    return {'ai_score': ai_score, 'overall': overall_score, 'gaps': gaps}

visibility = ai_mode_visibility(results, BRAND)

def save_daily_snapshot(results, visibility, output_file='ai_mode_tracking.json'):
    try:
        with open(output_file) as f:
            history = json.load(f)
    except FileNotFoundError:
        history = []
    snapshot = {
        'date': datetime.now().strftime('%Y-%m-%d'),
        'ai_score': visibility['ai_score'],
        'overall_score': visibility['overall'],
        'keywords_tracked': len(results),
        'ai_cited': sum(1 for r in results if r['in_ai_mode']),
        'details': results,
    }
    history.append(snapshot)
    with open(output_file, 'w') as f:
        json.dump(history, f, indent=2)
    # Trend analysis
    print(f'\n=== Trend ===')
    if len(history) >= 2:
        prev = history[-2]
        delta = snapshot['ai_score'] - prev['ai_score']
        direction = 'UP' if delta > 0 else 'DOWN' if delta < 0 else 'STABLE'
        print(f'  AI Score: {prev["ai_score"]:.0f} -> {snapshot["ai_score"]:.0f} ({direction} {abs(delta):.0f}pt)')
        print(f'  AI Citations: {prev["ai_cited"]} -> {snapshot["ai_cited"]}')
    else:
        print(f'  First snapshot saved. Run daily to track trends.')
    print(f'\n  Daily cost: ${len(results) * 0.005:.3f}')
    print(f'  Monthly cost: ${len(results) * 0.005 * 30:.2f}')

save_daily_snapshot(results, visibility)