How do I find the buying-intent comments on my Instagram and TikTok posts?

Short answer: export the comments with a comment scraper, then label each one by what it is — purchase intent, question, complaint, request, praise or spam — and filter. Social Comment Classifier does the labelling step: it takes the dataset from an Instagram, TikTok, Facebook or YouTube comment scraper and returns a type, a sentiment and a "needs reply" flag for every comment. It costs $0.50 per 1,000 comments, with no prompt writing and no LLM key.

"Price?", "link?" and "when is the restock?" are easy to miss under a few hundred emoji and "love this". Those are the comments closest to a sale, and the ones a brand most wants to answer first.

What you get

Real rows from a run on 2026-10-09 — public comments from the four comment scrapers listed below, with one custom label ("mentions shipping or delivery"):

Each comment gets one main type, a probability for every type (typeScores), positive / neutral / negative sentiment, and whether the brand or creator should answer. Rows keep the comment ID and post fields from the scraper (id, cid, postUrl, videoWebUrl, …), so you can jump back to the comment.

When a comment fits several types, one wins in a fixed order — spam, then purchase intent, complaint, question, request, praise. "Love it! Where can I buy it in the UK?" is purchase_intent, not praise.

Each run also saves a free summary by post: the comment-type mix, sentiment shares, the share of comments that need a reply, and up to 3 example comments for purchase intent, questions, complaints and requests.

How to set it up (5 minutes)

1. Export the comments. Run one of the common comment scrapers on Apify Store. Field detection was checked against real output of each: Instagram Comments Scraper, TikTok Comments Scraper, Facebook Comments Scraper, YouTube Comments Scraper.
2. Open the classifier on Apify: Social Comment Classifier.
3. Pass the scraper's dataset. The text, post title and ID fields are detected automatically. Add up to 10 yes/no labels of your own in plain language if you need them:

   {
     "datasetId": "YOUR_COMMENTS_DATASET_ID",
     "customLabels": ["mentions shipping or delivery"]
   }

To try it without a dataset, the form default classifies two pasted comments for about $0.001.

1. Filter the results. purchase_intent is your sales list. Needs reply = yes is your answer queue. Complaints and requests can go to support and the product team.
2. Schedule it with Apify Schedules — for example, classify new comments every morning.

For scale: 10,000 comments cost $5.00, and about 100 comments take 3–4 seconds. Apify's free $5 monthly credit covers about 10,000 comments. Comments without text are skipped and not charged.

You can also ask Claude, Cursor or Claude Code through the Apify MCP server — "Classify dataset abc123 from my TikTok comments run and list the unanswered questions." Setup is on the tool page.

How accurate is it?

On 100 hand-labelled public English comments that were not used for tuning, the main type matched 90% of the time and sentiment 87%. Complaints were found 97% of the time and questions 100%. Purchase intent was found 71% of the time, but that set had only 7 purchase-intent comments, so treat it as a rough number. Very short comments without context ("No", "Vibes") are the hardest.

English is measured. Other languages are accepted and usually work, but their accuracy has not been measured — treat them as beta. Automated labels can be wrong; check a sample before a big decision.

Other ways, and their limits

• Read the comments yourself. Fine for one post. It stops scaling once you have several posts a week, and the "link?" comments are buried in praise.
• Search for keywords like "price", "link" or "buy". Quick, but it misses "when is the restock?" and "do you ship to Canada?", and catches praise that happens to contain the word.
• Prompt a general chatbot. It can label a pasted batch, but you write and maintain the prompt and parse free text back into a table. This tool returns fixed labels with probabilities instead.

Independent tool — not affiliated with Meta (Instagram, Facebook), TikTok, Google (YouTube) or the authors of the scrapers listed above. The comments shown are public comments used as examples. More data tools for e-commerce and social listening: leoworks.kr/tools.