Content analysis in ChatGPT
Give the model the words, not the view count.
Engagement numbers tell you a post did well and nothing about why. The why is in the content itself — what was said in the video, and what people replied underneath. Both are text once they are fetched: transcripts turn video into something a model can read, and the comment endpoints return whole threads rather than the visible top few. With KonbiniAPI connected, ChatGPT reads the substance instead of the scoreboard.
- 1A KonbiniAPI accountThe free plan includes 100 credits, which is enough to run this a few times.
- 2ChatGPT connectedOAuth. Availability depends on your plan — Plus and Pro need developer mode turned on.
- 3A real targetA public account, post, or keyword to run it against — the placeholders below are not live values.
Run it.
Once ChatGPT is connected, send one of these. The agent decides which KonbiniAPI tools to call and returns the result in its answer — each successful call costs the same credits as the matching REST call.
Good for
- Find out what a high-performing video actually said
- Read a whole comment thread instead of the top few replies
- Summarize a discussion across a subreddit rather than one post
- Pull the arguments people make about a product, in their own words
Get the transcript of this TikTok video and summarize the argument it makes: [url].
Pull the comments on @[creator]'s most recent TikTok video, replies included, and tell me what people disagree about.
Read the top posts in r/[subreddit] this week and summarize what the community is discussing.
Fetch the comments on this Reddit post and group them by the position they take: [url].
Get the transcript of this LinkedIn video post and pull out the claims worth checking: [url].
Compare the comment sentiment on @[brand]'s last three Instagram posts. What changed?
Replace @[username] and [topic] before sending — they are placeholders, not live values.
Other ChatGPT workflows
- Creator researchAsk about an account and bring its public profile and recent posts into the conversation.
- Competitor analysisBring the competitor's actual posting history into the conversation, then reason about it.
- Influencer vettingOne message instead of six tabs and a spreadsheet.
- Brand monitoringCheck what a brand is publishing, and what people are saying back.
- Lead researchTurn a company name into what they do, how big they are, and what they are saying.
Run this in ChatGPT.
One API key works for MCP and the REST API. The free plan includes 100 credits and no credit card.