Pain-Point Radar
Find customer pain points in TikTok, YouTube, RedNote and Douyin posts
“Most users complain about suction” is what a summary gives you. How many posts is “most”? Name a category and the radar reads the top posts, groups what people complain about, want and hesitate over, and counts in code how many posts each point appears in, with a quote you can open. It also shows what buyers type into search boxes, and the searches no post answers.
What one run returns
Seven sections, every line counted
- Pain points: complaints and failures, each as “N of M posts” with quotes.
- Needs and selling points: what buyers look for, and what the posts praise (top posts can be sponsored).
- Purchase blockers: price, trust, fit, set-up.
- Products named: brands and models, by how many posts name them.
- Unanswered questions: what readers ask that no post answers.
- Search demand: search suggestions grouped into buy, worth it, compare, problems, price, scenario.
- Opportunities: searches that none of the read posts covers; and on a re-run, what changed since last time.
A live run
robot vacuum: what the top YouTube and TikTok reviews complain about
This category's radar is still being read. Finished ones are listed at the library (in Chinese).
For developers and agents
One call
curl -X POST https://linkdigest.dev/api/v1/radar \
-H "Authorization: Bearer $LINKDIGEST_API_KEY" -H "Content-Type: application/json" \
-d '{"topic": "robot vacuum", "market": "global", "size": "quick", "language": "en"}'
# then poll GET /api/v1/radar/{id} (usually 10-20 minutes); ?format=md for Markdownmarket: "global" reads YouTube and TikTok; "cn" reads RedNote and Douyin. Reports can be written in English, Chinese or Japanese; quotes stay in the post's own language with a translation. Agent skills on ClawHub: overseas-pain-points, xiaohongshu-pain-point-radar. Weekly public examples (in Chinese): overseas categories · China categories.
Questions
Before you run one
How do you identify customer pain points?
Read what customers write and say, not what you guess: reviews, posts and comments. Group the complaints that mean the same thing, count how many separate posts each one appears in, keep a quote for each, and compare them with what people type into search boxes. A pain point that shows up in many posts and in searches is worth acting on.
Where does the data come from?
The top public search results on each platform: for overseas buyers, YouTube and TikTok review videos (speech and on-screen text) plus Bing and YouTube search suggestions; for China, RedNote (Xiaohongshu) and Douyin posts, the top comments their pages show, plus Baidu, Taobao and Bing suggestions. No logins and no cookies.
Are the counts real?
Yes. A model only groups findings that say the same thing; the number of posts behind each group is counted by code from post numbers. Every quote was found word for word in its post, and posts about something else that shares the name are counted out and named.
Does it read comments, Amazon reviews or Reddit?
Top comments only where the platform's own page shows them (RedNote). YouTube asks servers to sign in before comments and we don't use cookies, so overseas runs read the videos themselves. Amazon reviews and Reddit are not read.
What does it cost?
A quick run reads 12 posts: about 41 credits for YouTube + TikTok and 47 for RedNote + Douyin (about $0.82–0.94 at the 250-credit pack). Posts already read cost nothing. Sign-up gives 10 free credits.