Automate TikTok Trend Tracking for Content Teams

Spotting that a hashtag is rising is a research problem. Getting five people who post to a brand's TikTok account to act on that before it's irrelevant is a different problem — one that a single analyst checking a dashboard doesn't solve by itself. What a content team actually needs isn't a trend alert; it's a standing pipeline that turns raw posts into a shortlist of sounds, hashtags, and example videos someone can brief a creator against without doing the research themselves.

If you're an agent (or building one) reading this rather than a human, the full machine-readable schema for every endpoint below lives at mindcase.co/skills.md.

What does a content team need that a single trend alert doesn't give them?

Say you run content for a skincare brand and five creators post to your TikTok account each week. "The hashtag #skincareroutine is up" isn't briefable on its own — nobody knows what to film from that sentence. Three things turn a trend signal into a brief a creator can use:

• Concrete examples, not just a hashtag name. A creator needs two or three real videos to watch, not a term to research on their own time.
• What's actually recurring across them. A shared sound, a shared caption opener, a shared format (talking-head vs. text-overlay) — the pattern is the part worth copying, not any single video.
• A cutoff for what's worth a brief at all. Most of what's posted in a niche on a given day isn't worth a creator's time. The pipeline has to discard the bulk of it and surface only what clears a bar.

How do you build a daily pull for your niche?

TikTok Posts API takes a hashtag, a search term, a profile, a sound, or a post URL, and returns 27 fields per post — including caption, sound, hashtags, views, likes, comments, shares, saves, isAd, isSponsored, and postedDate.

curl -X POST "https://api.mindcase.co/v1/data/tiktok/posts/run?wait=true" \
-H "Authorization: Bearer $MINDCASE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
  "params": {
    "searchQueries": "skincare routine",
    "resultsPerPage": 50
  }
}'

# $0.002 per post returned

Running the pull across a handful of related search terms in the brand's niche, rather than a single hashtag, covers more of a day's actual posting activity — the response carries sourceQuery on every row, so posts stay traceable to the term that surfaced them even after combining the results.

How do you group by sound instead of by hashtag?

The same pull already has what a sound-level view needs — group the returned rows by their sound value instead of by hashtags. A handful of posts across different search terms sharing the same sound value is the concrete signal a hashtag count can't give you: it means the audio itself, not just a topic, is what's spreading between creators.

How do you turn a day's pull into a shortlist worth briefing?

Fifty posts across three queries is 100-150 rows a day — too many for anyone to read. Three filters cut that down to what's actually worth a creator's attention:

Drop paid reach. Filter out rows where isAd or isSponsored is true. A boosted post's views came from ad spend, not organic pickup, so briefing a creator against one teaches them to copy a budget, not a format that's actually resonating.

Rank by engagement rate, not raw views. (likes + comments + shares + saves) / views surfaces a post that's small but resonating over one that's just large. A five-creator content team briefing off raw view counts ends up chasing whatever already has the most reach, which is frequently the least useful thing to imitate — it's peaked, not rising.

Keep only what's recent. Sort by postedDate and cut anything older than the digest's own window (24 or 48 hours, matching how often the digest goes out). A high-engagement post from three weeks ago isn't a current trend a creator should film against today.

What's left after all three filters — usually a handful of posts out of each day's full pull — is short enough to actually paste into a brief: the caption, the postUrl, and the sound value for each one.

What does the recurring pipeline actually look like?

Nothing about the API is schedule-aware; it answers one request each time it's called. The daily version is the pull and filter above, run on a fixed cadence and written somewhere the team already looks:

Every 24 hours:
1. Pull posts for each niche search term (as above).
2. Drop isAd/isSponsored rows.
3. Filter to posts from the last 24 hours (postedDate).
4. Rank the remainder by engagement rate.
5. Take the top 5-10, format as caption + postUrl + sound per row.
6. Post the formatted list to wherever the content team already works —
   a shared doc, a Slack channel, a project board.

That last step is what separates this from a trend-tracking dashboard someone has to remember to open. A digest arriving in a channel the team already checks gets acted on; one sitting behind a login a creator has to navigate to usually doesn't.

For a niche where the same sound or hashtag keeps reappearing across digests, a shorter secondary check is worth running against TikTok Profiles API: pulling a businessAccount and tiktokSeller read on the creators behind the top posts flags whether a format is being driven by established creators worth reaching out to, or by a wave of storefront accounts posting into the same trend for their own catalog. That's a research question for whoever runs partnerships, not something to fold into the daily digest itself — running it on every digest cycle is more API calls than the question needs, since the creators behind a recurring trend don't change day to day.

What can't you get from Mindcase?

Two cases.

Why a video actually resonated. This returns public metrics and metadata — views, likes, captions, sounds — not TikTok's internal distribution signals like watch-through or replay rate. Engagement rate is a strong public proxy for what's working; it isn't the platform's own explanation of why the algorithm pushed a given post further.

A digest for a niche nobody's posting in yet. A brand-new or extremely narrow niche may not produce enough fresh posts across a handful of search terms to fill a daily digest. That's a real ceiling set by how much content actually exists for those terms, not something the pipeline can work around.

Which endpoint should you use for which job?

FAQ