How to find TikTok trends and creators before they peak

The endpoint

Make the call

Put your key in FASTFETCHZ_KEY. The target below is a real one — it works as written.

curl "https://fastfetchz.com/v1/tiktok/search-hashtags-trending?target=space" \
  -H "X-API-Key: $FASTFETCHZ_KEY"

What comes back

Trimmed to the fields that matter. The full payload carries more.

{
  "success": true,
  "service": { "platform": "TikTok", "name": "Search / Hashtags / Trending" },
  "credits_charged": 1,
  "data": {
    "search_item_list": [
      {
        "aweme_info": {
          "aweme_id": "7674254062115949845",
          "desc": "Time dilation. #space #nasa #Astronomy #starship",
          "create_time": 1786801518,
          "statistics": {
            "play_count": 470645,
            "digg_count": 45978,
            "comment_count": 713,
            "share_count": 594,
            "collect_count": 3690
          },
          "author": {
            "unique_id": "stellamap.net",
            "nickname": "Stellamap",
            "follower_count": 128000
          },
          "music": { "title": "original sound - n0tslowed" },
          "share_url": "https://www.tiktok.com/@stellamap.net/video/7674254062115949845"
        }
      }
    ],
    "cursor": "30",
    "has_more": 1
  },
  "cursor": "30"
}

Getting more than one page

Search returns 30 videos per call. The page size is fixed at the source and not adjustable — but you can pass limit to have us trim the response to fewer results after the fetch. Trimming never changes the price; the upstream page was retrieved either way.

The cursor is a plain numeric offset — page one ends at cursor: "30", pass that back for videos 31–60. Stop when has_more comes back 0. Because the cursor is an offset into a live search index, results can shift between pages while you are paging; for a stable dataset, page through once and store aweme_id for everything you saw.

# Page 1 — 30 videos for the keyword
curl "https://fastfetchz.com/v1/tiktok/search-hashtags-trending?target=space" \
  -H "X-API-Key: $FASTFETCHZ_KEY"
# -> { "data": { "search_item_list": [ ... ], "has_more": 1 }, "cursor": "30" }

# Page 2 — pass the offset back
curl "https://fastfetchz.com/v1/tiktok/search-hashtags-trending?target=space&cursor=30" \
  -H "X-API-Key: $FASTFETCHZ_KEY"

What it costs

One keyword in, thirty videos out

The call takes one required parameter, target — the keyword to search. A page costs one credit and returns the 30 videos TikTok's search ranks highest for that keyword, which is a blend of recency and engagement, not a pure chronological feed. Search a phrase the way your audience would type it: a product category, a niche topic, a hashtag without the #.

Hashtags still come through — in each video's desc caption and in the cha_list field — so one keyword search doubles as a map of the surrounding hashtag vocabulary. Collect the tags from the top 60 results and you have the next three searches to run.

curl "https://fastfetchz.com/v1/tiktok/search-hashtags-trending?target=space" \
  -H "X-API-Key: $FASTFETCHZ_KEY"
# -> 30 videos, each with desc, statistics, author, music

Reading the raw response

Unlike most endpoints, this one returns TikTok's native schema, not our normalized profile shape. Three keys do almost all the work. statistics holds the numbers: play_count, digg_count (that is likes in TikTok's vocabulary), comment_count, share_count and collect_count (saves — often the best intent signal of the five).

author holds unique_id (the @handle), nickname and the author's follower_count — which is how you spot the trend before it peaks: a video with 470,000 plays from a 128,000-follower account is outperforming its audience by 3.7×, and that account belongs on your watchlist. music holds the sound's title, and share_url gives you the direct link to the video.

# jq: handle, plays, likes, saves, sound — per video
curl -s "https://fastfetchz.com/v1/tiktok/search-hashtags-trending?target=space" \
  -H "X-API-Key: $FASTFETCHZ_KEY" | jq '.data.search_item_list[] | {
    handle: .aweme_info.author.unique_id,
    followers: .aweme_info.author.follower_count,
    plays: .aweme_info.statistics.play_count,
    likes: .aweme_info.statistics.digg_count,
    saves: .aweme_info.statistics.collect_count,
    sound: .aweme_info.music.title,
    url: .aweme_info.share_url
  }'

Spotting the trend, not just the video

Group by music.title across a few pages. When the same sound shows up on many independently successful videos in your niche, that is the trend — a reusable format anyone can ride. The same goes for desc patterns: repeated hook phrasings across unrelated authors are a format, not a coincidence.

Rising versus peaked is a ratio you compute from fields already in the response: plays divided by the author's follower count. Search ranking already favors engagement, so a small account clearing the top 30 is the signal worth chasing. Store create_time with every result and a weekly run of the same keyword turns into a trend line you own.

From discovery to tracking

Discovery hands you handles; the tracking endpoints turn them into a dashboard. tiktok/profile gives you the account-level numbers in one call, and tiktok/videos-user-videos returns a creator's feed ten videos at a time — the same two calls behind our creator-tracking guide. tiktok/transcript converts any video's audio to text if you want to study the hooks, not just the numbers.

curl "https://fastfetchz.com/v1/tiktok/profile?target=stellamap.net" \
  -H "X-API-Key: $FASTFETCHZ_KEY"
# -> follower_count, following_count, post_count, like_count, biography

What you cannot get

There is no region filter and no publish-date filter on this route — the supplier endpoint takes a keyword and a page offset and nothing else, so freshness is something you judge from create_time rather than set. There is no creator-only search either: creators appear in results only as the authors of matching videos.

The 30-per-call page size is fixed at the source, and the cursor is an offset into a live index, so a deep page can repeat or skip a video if the rankings moved between your calls — dedupe on aweme_id when you store. Results are current search rankings, not a historical archive; a trend line comes from running the same search on a schedule and keeping what each run returned.

Endpoints you will probably want next