TikTok impulsa las tendencias de productos más rápido que cualquier otra plataforma: un producto puede pasar de ser desconocido a estar agotado en cuestión de días. La detección temprana de tendencias brinda a las marcas de comercio electrónico y a los dropshippers una ventaja crucial. Este tutorial crea un canal de detección de tendencias utilizando la API Scavio TikTok que monitorea los hashtags relacionados con productos, rastrea la velocidad de visualización e identifica productos emergentes antes de que alcancen su punto máximo. El sistema analiza las tasas de crecimiento de los hashtags, los patrones de adopción de los creadores y los índices de participación para generar puntuaciones de tendencias. Cada ejecución de detección cuesta entre 5 y 10 créditos (entre 0,025 y 0,05 dólares).
Requisitos previos
- Python 3.9+ instalado
- solicita biblioteca instalada
- Una clave API de Scavio de scavio.dev
- Categorías de productos o nichos a monitorear
Guia paso a paso
Paso 1: Definir categorías de seguimiento de productos
Configure las categorías de productos y los términos de búsqueda relacionados para monitorear. Cada categoría tiene varias palabras clave para captar diferentes ángulos de la misma tendencia.
Python
categories = {
'skincare': {
'keywords': ['skincare routine', 'skincare haul', 'skincare viral',
'skincare hack', 'skincare tiktok made me buy'],
'hashtags': ['skincareroutine', 'skincarehaul', 'tiktokmademebuyit']
},
'tech_gadgets': {
'keywords': ['tech gadget viral', 'amazon finds tech',
'tiktok gadget review', 'must have tech 2026'],
'hashtags': ['techfinds', 'amazontechfinds', 'gadgetreview']
},
'kitchen': {
'keywords': ['kitchen gadget tiktok', 'cooking hack viral',
'amazon kitchen finds', 'kitchen must have'],
'hashtags': ['kitchenhacks', 'kitchenfinds', 'cookingtiktok']
}
}
total_searches = sum(len(c['keywords']) + len(c['hashtags']) for c in categories.values())
print(f'{len(categories)} categories, {total_searches} total API calls')
print(f'Estimated cost per run: ${total_searches * 0.005:.2f}')Paso 2: Busque vídeos de productos de tendencia
Para cada palabra clave, busque TikTok y recopile datos de video. Realice un seguimiento de las reproducciones, los me gusta y los tiempos de creación para calcular métricas de velocidad.
Python
import requests, os, time
API_KEY = os.environ['SCAVIO_API_KEY']
TIKTOK_URL = 'https://api.scavio.dev/api/v1/tiktok'
def search_tiktok(keyword: str, count: int = 20) -> list:
resp = requests.post(f'{TIKTOK_URL}/search/videos',
headers={'Authorization': f'Bearer {API_KEY}',
'Content-Type': 'application/json'},
json={'keyword': keyword, 'count': count, 'cursor': 0})
resp.raise_for_status()
videos = resp.json().get('data', {}).get('videos', [])
return [{
'id': v.get('id', ''),
'author': v.get('author', {}).get('uniqueId', ''),
'desc': v.get('desc', ''),
'plays': v.get('stats', {}).get('playCount', 0),
'likes': v.get('stats', {}).get('diggCount', 0),
'comments': v.get('stats', {}).get('commentCount', 0),
'shares': v.get('stats', {}).get('shareCount', 0),
'create_time': v.get('createTime', 0),
'keyword': keyword
} for v in videos]Paso 3: Extraer nombres de productos de descripciones de vídeos
Analice las descripciones de los videos para identificar marcas y nombres de productos específicos. Los vídeos suelen mencionar el nombre exacto del producto, la marca o el ASIN de Amazon.
Python
import re
from collections import Counter
def extract_products(videos: list) -> Counter:
"""Extract product mentions from video descriptions."""
product_counter = Counter()
# Common patterns: "Product Name" in caps, @brand mentions, #productname
for v in videos:
desc = v.get('desc', '')
# Hashtag products (e.g., #CeraVe #StanleyCup)
hashtags = re.findall(r'#(\w+)', desc)
for tag in hashtags:
# Filter out generic tags
if len(tag) > 3 and tag.lower() not in {
'fyp', 'foryou', 'viral', 'trending', 'tiktok',
'skincare', 'review', 'haul', 'musthave'}:
product_counter[tag.lower()] += 1
# Brand mentions (capitalized words that look like brand names)
brands = re.findall(r'\b([A-Z][a-zA-Z]+(?:\s[A-Z][a-zA-Z]+)?)\b', desc)
for brand in brands:
if len(brand) > 3 and brand.lower() not in {'this', 'that', 'the', 'with'}:
product_counter[brand.lower()] += 1
return product_counter
# Example:
products = extract_products(search_tiktok('skincare viral'))
for product, count in products.most_common(10):
print(f' {product}: {count} mentions')Paso 4: Calcular puntuaciones de velocidad de tendencia
Califique cada producto según la rapidez con la que crecen sus menciones. La alta velocidad de visualización y muchos creadores únicos indican un producto de tendencia.
Python
import statistics
from datetime import datetime
def calculate_trend_score(product: str, videos: list) -> dict:
relevant = [v for v in videos if product in v.get('desc', '').lower()]
if len(relevant) < 2:
return {'product': product, 'trend_score': 0, 'reason': 'Too few videos'}
total_plays = sum(v['plays'] for v in relevant)
total_engagement = sum(v['likes'] + v['comments'] + v['shares'] for v in relevant)
unique_creators = len(set(v['author'] for v in relevant))
# Recency: weight recent videos more heavily
now = time.time()
recency_scores = []
for v in relevant:
age_days = (now - v['create_time']) / 86400 if v['create_time'] > 0 else 30
recency_scores.append(max(0, 1 - (age_days / 30))) # 0-1, 1=today
avg_recency = statistics.mean(recency_scores)
# Composite trend score
play_score = min(total_plays / 100000, 40) # up to 40 points
creator_score = min(unique_creators * 5, 30) # up to 30 points
recency_score = avg_recency * 30 # up to 30 points
total_score = round(play_score + creator_score + recency_score, 1)
return {
'product': product,
'trend_score': total_score,
'total_plays': total_plays,
'unique_creators': unique_creators,
'video_count': len(relevant),
'avg_recency': round(avg_recency, 2)
}Paso 5: Ejecutar todo el proceso de detección de tendencias
Combine todos los pasos en un único proceso que escanea categorías, extrae productos, califica tendencias y genera un informe clasificado.
Python
def detect_trends(categories: dict) -> list:
all_videos = []
credits_used = 0
for category, config in categories.items():
print(f'Scanning {category}...')
for keyword in config['keywords']:
videos = search_tiktok(keyword, count=20)
all_videos.extend(videos)
credits_used += 1
time.sleep(0.3)
# Extract and score products
product_counts = extract_products(all_videos)
trends = []
for product, count in product_counts.most_common(20):
if count >= 3: # minimum mention threshold
score = calculate_trend_score(product, all_videos)
if score['trend_score'] > 10:
trends.append(score)
trends.sort(key=lambda t: t['trend_score'], reverse=True)
print(f'\nDetected {len(trends)} trending products')
print(f'Credits used: {credits_used} (${credits_used * 0.005:.2f})')
for t in trends[:10]:
emoji_bar = '#' * int(t['trend_score'] / 5)
print(f' [{t["trend_score"]:5.1f}] {t["product"]}: '
f'{t["total_plays"]:,} plays, {t["unique_creators"]} creators '
f'{emoji_bar}')
return trends
trends = detect_trends(categories)Ejemplo en Python
Python
import os, requests, time, re
from collections import Counter
API_KEY = os.environ['SCAVIO_API_KEY']
TT = 'https://api.scavio.dev/api/v1/tiktok'
def search(keyword, count=20):
resp = requests.post(f'{TT}/search/videos',
headers={'Authorization': f'Bearer {API_KEY}', 'Content-Type': 'application/json'},
json={'keyword': keyword, 'count': count, 'cursor': 0})
return resp.json().get('data', {}).get('videos', [])
def detect_trends(keywords):
all_videos = []
for kw in keywords:
all_videos.extend(search(kw))
time.sleep(0.3)
products = Counter()
for v in all_videos:
for tag in re.findall(r'#(\w{4,})', v.get('desc', '')):
if tag.lower() not in {'fyp', 'foryou', 'viral', 'trending'}:
products[tag.lower()] += 1
print(f'Scanned {len(all_videos)} videos, found {len(products)} products')
for product, count in products.most_common(10):
print(f' {product}: {count} mentions')
return products
detect_trends(['skincare viral', 'amazon finds 2026', 'tiktok made me buy'])Ejemplo en JavaScript
JavaScript
const API_KEY = process.env.SCAVIO_API_KEY;
const TT = 'https://api.scavio.dev/api/v1/tiktok';
async function searchTikTok(keyword) {
const resp = await fetch(`${TT}/search/videos`, {
method: 'POST',
headers: { 'Authorization': `Bearer ${API_KEY}`, 'Content-Type': 'application/json' },
body: JSON.stringify({ keyword, count: 20, cursor: 0 })
});
return (await resp.json()).data?.videos || [];
}
async function detectTrends(keywords) {
const products = {};
for (const kw of keywords) {
const videos = await searchTikTok(kw);
videos.forEach(v => {
const tags = (v.desc || '').match(/#(\w{4,})/g) || [];
tags.forEach(t => {
const tag = t.slice(1).toLowerCase();
if (!['fyp', 'foryou', 'viral'].includes(tag)) {
products[tag] = (products[tag] || 0) + 1;
}
});
});
}
Object.entries(products).sort((a, b) => b[1] - a[1]).slice(0, 10)
.forEach(([p, c]) => console.log(` ${p}: ${c} mentions`));
}
detectTrends(['skincare viral', 'amazon finds 2026']);Salida esperada
JSON
Scanning skincare...
Scanning tech_gadgets...
Scanning kitchen...
Detected 8 trending products
Credits used: 12 ($0.06)
[ 78.5] cerave: 1,234,000 plays, 15 creators ################
[ 65.2] stanleycup: 890,000 plays, 12 creators #############
[ 52.1] dysonairwrap: 567,000 plays, 8 creators ###########
[ 45.8] theordinary: 445,000 plays, 9 creators #########
[ 38.4] airfryer: 334,000 plays, 7 creators ########
[ 31.0] laneige: 234,000 plays, 6 creators ######
[ 24.5] owala: 178,000 plays, 5 creators #####
[ 18.2] hexclad: 123,000 plays, 4 creators ####