> ## Documentation Index
> Fetch the complete documentation index at: https://monid.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# Find TikTok Videos Outperforming a Creator's Baseline

> Use Monid to surface videos beating each creator's own median, with sample sizes, baselines, and testable hypotheses.

Give your agent a niche, country, and budget. It uses Monid to discover recent posts, calculate creator-relative baselines, and return outlier videos worth studying.

<Tip>
  **Example question:** Which videos are outperforming for small specialty-coffee creators this month?
</Tip>

## Try it yourself (give it to your agent)

```text Prompt theme={null}
Use Monid to find unusually strong TikTok videos from specialty-coffee creators with fewer than 100,000 followers, posted in the last 30 days.

Ask which country and language to focus on, and whether I care most about brewing tutorials, equipment reviews, or cafe content. My budget is $2. Inspect prices, cap the initial discovery set at 50 videos, and compare no more than five shortlisted creators initially.

For each shortlisted creator, retrieve a bounded sample of recent posts and compare each candidate video's views with that creator's median. Show sample size, post age, raw views, baseline, and outlier ratio. Do not treat a single snapshot as evidence of accelerating growth.

Return five videos worth studying, their source links, and testable ideas about the hook or format. Treat explanations as hypotheses, not proof of why a video succeeded. Report missing metrics, collection limits, and actual spend.
```

## Potential tools

* [**TikTok**](https://monid.ai/tools/tiktok): Apify's `/apidojo/tiktok-scraper` supports keyword/profile collection.
* **TikHub**: `/api/v1/tiktok/web/fetch_user_post` can retrieve creator history when the correct account identifier is available.

The agent calculates creator-relative baselines and outliers.

## Example output

<Note>Names, counts, and details below are fictional. Real responses include actual sources, costs, and generated files.</Note>

> **The strongest outlier was a side-by-side grinder demonstration.**
>
> * **Views:** 84,000
> * **Creator's recent median:** 7,000
> * **Outlier ratio:** 12× baseline across 20 sampled posts
>
> A second standout was a *"fix this sour espresso"* video. Its opening shows the failed result **before** explaining the adjustment, giving viewers a clear reason to continue.
>
> **Suggested experiment:** Test a visible before/after result in the first few seconds. That is a **hypothesis** based on these examples, not proof that the opening caused their performance.
>
> All five selections include links and baseline calculations. I could **not** assess growth velocity because no earlier snapshots were available.