Research Douyin source indices with explicit windows, identities and prices.
Enable Index research
Enable Index on your Key and add packs=index to MCP. All three tools currently support Douyin. Each mode is an explicit call; date discovery, identity discovery and pagination are never hidden extra calls.
Find a category and supported dates
content_trends modes: categories, dates, keywords, keyword_items, topics, publish, interaction and consumption. Start with categories and dates, then use their values in your research. Keyword/topic periods are 1, 3 or 7 days; a 7-day period must end on Sunday. topics accepts rank=index|rise.
Trend windows include both endpoints and contain 1–90 calendar days. Dates cannot be in the future. Missing dates and missing values stay missing. Source rate series keep native_units_unspecified; they are not audience counts or standardized percentages.
REQUEST TEMPLATE
{
"platform": "douyin",
"mode": "categories"
}
Discover a brand, then inspect its window
brand_radar modes: discover, availability, radar, trend, cycle and weekly_rank. Supply the exact brand name returned by discover; availability returns source-supported dates where present. Other modes require start_date and end_date. weekly_rank requires exactly 7 days ending on Saturday.
Brand indices retain native_index units. A weekly rank is the requested brand's category position, not a complete leaderboard. Native ratios may be negative. Different units and windows must not be merged into a single score.
Use the same public creator identity
creator_comparison discover accepts query and count=1–20 (default 5). Each returned id is a public Douyin sec_uid that can continue into compare, metrics, user_profile or user_posts. Unresolvable candidates are omitted with a warning; source identifiers that cannot continue are not offered as public IDs.
compare accepts 2–5 distinct users, each a sec_uid or canonical https://www.douyin.com/user/<sec_uid> URL, and days=7|30 (default 7). It internally reads each public profile once, then makes one comparison request. The whole call costs 0.5/0.6/0.8/0.9 credit for 2/3/4/5 people. It returns window summaries, not daily curves. Source window changes and averages may be negative; cumulative counts remain nonnegative.
metrics accepts one user and returns the source's week and month averages and ratios together. It has no days parameter and costs 0.3 credit including its one profile lookup. compare and metrics have a shared 25-second deadline across all steps; discover uses 8 seconds and costs 0.2 credit. All workflow failures are free. A timeout does not authorize automatic retries of already dispatched work.
API call prices
A mode is billed once, including its documented identity lookups. Replays reuse the original result and price. Index scope does not grant the separate Core user tools; those require their own authorization.