Pass source amazon for a shopper-phrase search series, or type Amazon Best Sellers Top Rated for the live listing board. The two strings are not interchangeable.
Live data as of 2026-09-07
An Amazon search vs Amazon Best Sellers comparison on Trends MCP is two different tools. Source amazon on get_growth or get_time_series returns a monthly 0-100 search-interest series for a shopper phrase, with absolute volume when the pipeline has it. Type Amazon Best Sellers Top Rated on get_top_trends returns ranked listing titles and takes no keyword. From the September 7, 2026 pull (as_of 2026-09-07T04:01:37 UTC), rank 1 was an Amazon Basics AA battery pack whose amazon 30D volume was 6,869, while Crocs at rank 4 printed 6,697,760. 7D and 1W collapsed on that monthly grain. Swapping the two strings is a 400. The free plan includes 100 requests per month.
The September 7, 2026 Top Rated board is the split that still matters next month: live rank is not a search-volume ranking. Omitting limit returned limit 25, count 25, offset 0, as_of_ts 2026-09-07T04:01:37.991749+00:00. Rank 1 was an Amazon Basics 48-pack AA listing. Rank 4 was the Crocs Classic Clog listing. Follow-up get_growth calls on source amazon used recent_date 2026-08-31:
| Board rank | Shopper keyword | 30D index | 30D volume | 30D change |
|---|---|---|---|---|
| 1 | amazon basics aa batteries | 18.1 from 15.0 | 6,869 from 5,688 | +20.67% |
| 3 | Verity | 89.8 from 28.1 | 397,753 from 124,450 | +219.57% |
| 4 | crocs | 38.3 from 52.6 | 6,697,760 from 9,187,782 | -27.19% |
| 7 | essence lash princess | 30.8 from 27.5 | 41,263 from 36,873 | +12.0% |
| 12 | Project Hail Mary | 80.2 from 100.0 | 3,156,729 from 3,935,662 | -19.8% |
| 13 | Tide PODS | 100.0 from 77.9 | 867,974 from 676,024 | +28.37% |
Crocs held about 975 times the 30D volume of the rank 1 battery pack, and still sat three places lower. Verity at rank 3 jumped +219.57% in a month on 397,753 volume. Tide PODS sat at index 100.0 (its own peak) at rank 13. Project Hail Mary at rank 12 had already rolled off a 100.0 print on 2026-07-31, while its 12M window was +1062.32% from 6.9 on 2025-08-31 (volume 3,156,729 from 270,111).
Ranks 10 and 11 were the same string, blink plus plan with monthly auto-renewal. Source amazon on blink plus returned 34 points, 30D volume 1,066, and 12M -74.46% (46.2 to 11.8). Duplicate titles happen on this board. Treat them as one listing, not two demand events.
sort=rank_change with window=7d and limit=10 still used the same as_of_ts, plus baseline_ts 2026-08-31T04:01:36.214924+00:00. The first mover sat at rank 161 (prev_rank 180, rank_change 19). Dungeon Crawler Carl appeared three times (ranks 80, 80, 81). Mover ranks routinely sit past the default 25-row window.
Two strings, two tools. REST uses mode get_growth or get_top_trends. MCP uses the matching tool names. The keyword series page is Amazon search trends. The live board page is the Amazon Best Sellers API. Window defaults live on get_growth. Pagination and limit live on get_top_trends.
{"mode": "get_top_trends", "type": "Amazon Best Sellers Top Rated"}
{"mode": "get_growth", "source": "amazon", "keyword": "crocs", "percent_growth": ["30D", "1M", "3M", "12M", "YTD"]}
{"mode": "get_growth", "source": "amazon", "keyword": "air fryer"}
{"mode": "get_growth", "source": "amazon, google search", "keyword": "air fryer", "percent_growth": ["30D", "12M"]}
{"mode": "get_top_trends", "type": "Amazon Best Sellers by Category", "category": "Toys Games"}
The third call omitted percent_growth. The response still used period 12M, recent_value 54.4, baseline_value 59.5, growth -8.57, recent_volume 5,882,739. That matches the 12M row inside the multi-window air fryer block.
The comma-separated air fryer call returned sources_requested ["amazon", "google search"], sources_successful 2, and trend_analysis.uptrend_ratio 1/2. Amazon 30D was +5.63% (51.5 to 54.4) on recent_date 2026-08-31 with 50 points. Google Search 30D was -5.26% (57.0 to 54.0) on recent_date 2026-09-05 with 261 weekly points and volume_estimated true. Same preset, different close dates, opposite signs. Read source_results before joining those percentages.
get_time_series on air fryer + amazon returned 50 monthly rows from 2022-07-31 (index 67.1, volume 7,261,799) to 2026-08-31 (index 54.4, volume 5,882,739). Each row carried date, value, volume, keyword, and datatype amazon. Index 100.0 landed on 2025-11-30 with volume 10,823,067. That November peak is why 12M is negative while 30D is slightly up.
percent_growth 7D and 1W on crocs both returned status error, error collapsed_range, and the message that recent and baseline resolved to the same data point. air fryer did the same on 7D. The series is monthly. There is no distinct week inside one closing month.
1M is the alias that works. On air fryer, 1M and 30D both printed recent_date 2026-08-31, baseline_date 2026-07-31, recent_value 54.4, baseline_value 51.5, growth 5.63, recent_volume 5,882,739. On amazon basics aa batteries, both presets printed +20.67% from 15.0 to 18.1. The schema lists 1M as a preset. It is not a daily window on this source.
Successful longer windows on crocs used the same close: 3M -39.3% from 63.1 on 2026-05-31 (volume 6,697,760 from 11,033,171), 12M -27.46% from 52.8 on 2025-08-31, YTD -33.04% from 57.2 on 2025-12-31. Volume growth tracked the index within 0.2 points on those rows.
blink plus is the short-history case: 34 points, not 50. 12M still computed (-74.46%). A brand-new listing can miss entirely. The Crocs board title as keyword is that miss.
type amazon on get_top_trends returned HTTP 400 error invalid_request and listed 24 valid feed names, starting at Amazon Best Sellers Top Rated and Amazon Best Sellers by Category. source Amazon Best Sellers Top Rated on get_growth returned HTTP 400 error invalid_source, echoed 'amazon best sellers top rated' is not a valid source., and listed 16 keyword sources plus all (google search through python, then all). get_time_series on amazon, google search returned invalid_source too, with the 16 names and no all.
Lowercase amazon best sellers top rated as type still returned HTTP 200 and echoed Amazon Best Sellers Top Rated with the same as_of_ts and 25 rows. The feed label is not case-sensitive on this pull. The keyword source still is: pass amazon, not the board title.
REST POST https://api.trendsmcp.ai/api with mode get_top_trends and no type returned HTTP 200 with envelope statusCode 400, error invalid_request, and the same 24-type list. That omit-type envelope is the sibling of the type amazon miss. Both need the exact Top Rated or by-category label.
Passing type Amazon Best Sellers by Category with category Toys Games returned 25 Toys Games rows, same as_of_ts as Top Rated, rank 1 Bezente Assorted Latex Rainbow Balloons. That aisle list is a 25-deep chart inside one official category string.
Omitting category still returned HTTP 200, count 25, same stamp. Every row was rank 1, each tagged with a different aisle: Kitchen Dining (Stanley Quencher 30 oz), Electronics (blink plus), Clothing Shoes Jewelry (Hanes hoodie), Toys Games (the same balloon pack), Grocery Gourmet Food (CELSIUS 12-pack), Video Games (Roblox 1,000 Robux). The mixed call is one winner per aisle, not a 25-deep Toys Games page. Official category names include Toys Games, Kitchen Dining, and Electronics. Skip category only when the job is "who leads each aisle," then filter in code.
get_top_trends MCP pricing is 1 credit per 10 requested rows, max 10 credits per call. REST counts 1 request per call. Free boards still cap at the top 10. The 25-row pulls on this page are the paid default, which is also what the handler returns when limit is omitted on a paid key.
FAQ