TikTok Trend Product Radar: spot winning dropshipping products twice a week, automatically
An n8n workflow that scrapes trending TikTok product videos twice a week, has Claude score each product for saturation and audience fit, then logs everything to Google Sheets and pings your Telegram with the top 5.
- 1
Schedule the radar
Add a
Schedule Triggernode set to Monday and Thursday at 09:00. Twice a week is the sweet spot: TikTok trends move fast, but daily runs mostly return the same videos and burn Apify credits. - 2
Scrape trending product videos with Apify
Add an
HTTP Requestnode calling Apify's TikTok Scraper actor (clockworks/tiktok-scraper). Request the last 3 days of videos for hashtags#tiktokmademebuyitand#amazonfinds, limited to 50 results, returning video URL, views, likes, and the caption text. Paste your Apify token into an n8n credential — never into the node itself. - 3
Filter out the noise
Add a
Codenode that keeps only videos above 100k views, computes alikes/viewsengagement ratio, and de-duplicates videos that showcase the same product. You should end up with 10–25 candidate videos per run. - 4
Let Claude score every product
Add an
HTTP Requestnode to the Claude API (/v1/messages). System prompt: extract distinct products from the captions and stats, and for each return JSON withproduct_name,category,why_trending(one sentence),target_audience, andsaturation(low / medium / high). Instruct it to answer with JSON only — no prose — so the next node can parse it directly. - 5
Log everything to Google Sheets
Add a
Google Sheetsnode in append mode. One row per product: date, product name, category, saturation, engagement ratio, source video URL. Within a month this sheet becomes your private trend database — and the raw material for content if you publish your own picks. - 6
Get the top 5 in Telegram
Add a
Telegramnode that sends the 5 highest-scoring products (low saturation + high engagement first) as a short digest to your channel or DM. Product research now takes zero minutes of your week — you just read the message and pick what to test.