Vetting Influencer Follower Quality via TikTok API

Checking follower quality via TikTok API: pull a follower sample, measure engagement rates against follower count, and detect bot patterns from account characteristics. At $0.005/credit and 3-4 API calls per creator, vetting costs roughly $0.015-0.02 per creator — far cheaper than a paid influencer platform subscription.

The Core Signals

Bot followers have identifiable patterns. You do not need to check every follower — a sample of 100-200 is statistically sufficient for most vetting decisions:

1. Engagement rate vs follower count: legitimate accounts in most niches have 2-8% engagement rate (likes/comments/views relative to followers). Accounts with 500k followers and 0.1% engagement have likely purchased followers.

2. Follower account characteristics: bot accounts tend to have no profile picture, no bio, zero or very few videos, and follow thousands of accounts while having few followers themselves.

3. Comment quality: genuine comments reference video content. Bot comments are generic ("great video!", "love this", strings of emojis — none present here).

API Call Sequence

import requests

BASE = "https://api.scavio.dev/api/v1/tiktok"
HEADERS = {"Authorization": f"Bearer {API_KEY}"}

def vet_creator(username: str) -> dict:
    # 1. Get profile (1 credit)
    profile = requests.post(
        f"{BASE}/user_info",
        headers=HEADERS,
        json={"username": username}
    ).json()

    follower_count = profile.get("follower_count", 0)

    # 2. Get recent videos for engagement rate (1 credit)
    videos = requests.post(
        f"{BASE}/user_videos",
        headers=HEADERS,
        json={"username": username, "count": 10}
    ).json().get("videos", [])

    # 3. Get follower sample (1 credit)
    followers = requests.post(
        f"{BASE}/user_followers",
        headers=HEADERS,
        json={"username": username, "count": 100}
    ).json().get("followers", [])

    return analyze_quality(profile, videos, followers)

def analyze_quality(profile: dict, videos: list, followers: list) -> dict:
    # Engagement rate
    follower_count = profile.get("follower_count", 1)
    if videos:
        avg_likes = sum(v.get("digg_count", 0) for v in videos) / len(videos)
        engagement_rate = avg_likes / follower_count
    else:
        engagement_rate = 0

    # Follower quality signals
    bot_indicators = 0
    for f in followers:
        score = 0
        if not f.get("avatar_thumb"): score += 1
        if not f.get("signature"): score += 1
        if f.get("aweme_count", 0) == 0: score += 1
        if f.get("following_count", 0) > 2000 and f.get("follower_count", 0) < 100:
            score += 2
        if score >= 3:
            bot_indicators += 1

    bot_rate = bot_indicators / len(followers) if followers else 0

    return {
        "username": profile.get("unique_id"),
        "followers": follower_count,
        "engagement_rate_pct": round(engagement_rate * 100, 2),
        "estimated_bot_rate_pct": round(bot_rate * 100, 1),
        "recommendation": "proceed" if engagement_rate > 0.02 and bot_rate < 0.3 else "review"
    }

3 credits total per creator = $0.015 at $0.005/credit.

Interpreting Engagement Rate by Category

Engagement rate benchmarks vary significantly by niche and follower count. Nano-influencers (10k-100k) typically have higher engagement rates than mega-influencers (1M+). Do not apply a single threshold:

Accounts outside the expected range for their follower tier warrant closer inspection.

What This Analysis Cannot Detect

The API-based approach has limits:

• Purchased comments: sophisticated operations sell high-quality comments, not just likes
• Follow-unfollow manipulation: accounts that gain followers through follow-unfollow cycles have real followers with inflated counts but legitimate account profiles
• Regional bot networks: some bot farms use real device accounts with actual activity, making profile-based detection unreliable

For high-budget influencer partnerships, supplement API analysis with manual review of the most recent 5-10 comment sections. This catches quality-comment bot farms that API signals miss.

When to Use a Paid Platform Instead

Paid influencer platforms (Modash, Heepsy, Upfluence at $150-500+/month) add audience demographic data, historical growth charts, and fraud probability scores from proprietary training data. For agencies managing dozens of influencer relationships monthly, the platform cost is justified.

For brands vetting 10-20 creators per quarter, the API approach at $0.02/creator is dramatically cheaper and sufficient for basic quality checks.