Amazon search interest is a shopper signal. TikTok interest is a short-form signal. They share a 0-100 on this API and they do not share a window grid. get_growth with source amazon, tiktok is the join. body is a JSON string and must be parsed a second time. Docs: the API reference. Amazon notes: the Amazon guide. TikTok notes: the TikTok guide. Caps: pricing.

Search demand versus hashtag demand

POST https://api.trendsapi.ai/api
Authorization: Bearer <api_key>
Content-Type: application/json

{"mode":"get_growth","source":"amazon, tiktok","keyword":"air fryer","percent_growth":["7D","14D","1M"]}

A live pull on 2026-08-27 kept Amazon on air fryer and TikTok on airfryer. Social spelling and retail spelling diverge. Store search_term per source.

Keyword format is a phrase on both

No bundle ID. No ASIN requirement. No #. If the inner Amazon object is empty, try a broader product phrase. If TikTok is empty, drop spaces the way the live pull did, or try the closed-up form first.

Best-seller rank is not this comparison

{"mode":"get_top_trends","type":"Amazon Best Sellers Top Rated","limit":25}

{"mode":"get_top_trends","type":"TikTok Trending Hashtags","limit":25}

Those boards have no keyword. They answer "what is on the list now." They do not answer "did Amazon search fall while TikTok rose for this phrase." Keep them off this chart.

A live pair from this session

Amazon rejected 7D and 14D (collapsed_range). TikTok accepted both (7D +43.18 percent, 14D +75.81 percent). The overlapping honest window was 1M:

  • amazon: 51.5 vs 57.1, -9.81 percent, volume 5,577,980 vs 6,178,129, recent_date 2026-07-31
  • tiktok: 37.8 vs 31.2, +21.15 percent, volume 3,096,263 vs 2,845,623, recent_date 2026-08-25

volume_weighted_growth on that call was -0.50 percent, Amazon 64 percent of blended volume, TikTok 36 percent. The blend hid the sign split. Report both sources, then the blend, never the blend alone. The product MCP server at https://api.trendsapi.ai/mcp uses the same comma-separated source list.