Launch timing is a two-date question. get_growth with percent_growth ["3M","12M"] is the review packet. POST https://api.trendsapi.ai/api. A year window that is up and a quarter window that is down means the product is still larger than last August and already off the spring peak. Shopping notes sit next to search on the Google Trends guide. Preset names sit on the API reference. Read results only after decoding body.
The late-ship pattern
Launch packet dated 27 Aug 2026 for air fryer on google search: score 56. Year +16.67%. Quarter −18.84%. The 1M print (+5.66%) is a small recover from July, not a new cycle. A memo that quotes only the year still ships late.
google shopping for the same phrase: 8 on 22 Aug 2026. 7D and 1M both +33.33% (8 vs 6). 3M −84.31% (8 vs 51). 12M +100% (8 vs 4). Shopping is a thin score with a violent seasonal range. Treat the 3M crash as "last May was the event." Treat the 7D lift as "shoppers are back in the aisle this week."
TikTok cannot gate a year
POST https://api.trendsapi.ai/api
Authorization: Bearer <api_key>
Content-Type: application/json
{"mode":"get_growth","source":"google search, google shopping, tiktok, amazon","keyword":"air fryer","percent_growth":["7D","1M","3M","12M"]}
TikTok stored airfryer, 37.8 on 25 Aug 2026, +43.18% on 7D, +21.15% on 1M, then date_out_of_range on 3M and 12M. A year-gate that requires TikTok will fail closed. Amazon 7D was collapsed_range. Amazon 1M was −9.81% (51.5 vs 57.1 on 31 Jul 2026). Do not inbound on a missing Amazon week. Do inbound only after search 3M, shopping 3M, and Amazon 1M sit on the same slide.
get_time_series on google search is the trough chart. Find the week the score last bottomed, then set the ship date after that week, not on the 7D print. Amazon notes: the Amazon guide.