Amazon product search growth API

What get_growth returns for Amazon keywords

A single get_growth call on source: "amazon" compares two points on the weekly Amazon product search curve and returns a percentage change for each requested window. The response includes normalized values (0-100 scale), direction, recent and baseline dates, and absolute volume fields when the pipeline has them.

For broader context on Amazon product signals, see Amazon product search volume data for AI agents.

Request shape

{
  "source": "amazon",
  "keyword": "collagen supplements",
  "percent_growth": ["3M", "12M", "YTD"]
}

MCP agents call the same parameters through get_growth. REST clients POST to https://api.trendsmcp.ai/api with a Bearer token. One keyword and one source per request; stack multiple periods in percent_growth without multiplying the request count.

Live example: collagen supplements (June 2026)

Trends MCP API snapshot, June 29 2026:

The 3M window looks modest. The 12M window shows demand roughly doubled year over year. That split is common in supplement categories where a Q1 push lifts the trailing-year baseline while the most recent quarter cools slightly. A buyer who only checks 3M might under-order; one who checks 12M sees durable expansion.

The weekly series behind this keyword shows a May 2023 spike to 569,758 estimated weekly searches, then a long normalization period before climbing again through early 2026. Growth percentages are point-to-point; they do not average the whole curve. Pair growth calls with get_trends when seasonality could distort a single window.

Live example: air fryer (June 2026)

Same API snapshot:

Air fryer is a mature category with millions of weekly searches. Here the 3M and YTD windows both read negative while 12M is slightly positive. That pattern often appears after a holiday peak: YTD compares against a high December baseline, so a post-Q4 slide shows up as a steep YTD decline even when annual demand is flat.

For listing optimization workflows, see Amazon search trends.

Response fields worth logging

Each period in results carries:

• period: the preset label or custom name
• growth: normalized point-to-point percentage
• direction: increase or decrease
• recent_date and baseline_date: the two weekly anchors
• recent_value and baseline_value: normalized 0-100 scores at those dates
• volume_available, recent_volume, baseline_volume, volume_growth: absolute demand when present

Store all periods from one call together. Dashboards that plot only growth without dates will mislabel a May-to-February comparison as "last quarter" when the API used specific weekly endpoints.

Custom date windows for launch tracking

Preset strings cover most FBA research. Product launches sometimes need an exact window:

{
  "source": "amazon",
  "keyword": "portable blender",
  "percent_growth": [
    {
      "name": "post_launch_90d",
      "recent": "2026-05-31",
      "baseline": "2026-02-28"
    },
    "12M"
  ]
}

Custom objects accept any valid weekly dates in the series. Name them for downstream reporting so a Make or n8n workflow can route alerts by label.

Which periods to pair for inventory decisions

Short windows (3M, 30D) catch reorder urgency and competitor promo effects. Medium windows (6M, 12M) separate fads from categories worth a second SKU. Calendar windows (YTD, QTD, MTD) align with finance reporting but punish seasonal categories after their peak month.

A practical default for private-label screening: request ["3M", "12M"] on the candidate keyword, then ["3M", "12M"] on two adjacent category terms. If the candidate beats its category on 12M but trails on 3M, dig into the weekly series before killing the listing.

Python example

import requests

resp = requests.post(
    "https://api.trendsmcp.ai/api",
    headers={"Authorization": "Bearer YOUR_API_KEY"},
    json={
        "source": "amazon",
        "keyword": "collagen supplements",
        "percent_growth": ["3M", "12M", "YTD"],
    },
)
payload = resp.json()
for row in payload["results"]:
    print(row["period"], row["growth"], row.get("volume_growth"))

MCP prompt pattern

Agents route more reliably when the source is explicit:

"Using TrendsMCP, show 3M, 12M, and YTD Amazon search growth for collagen supplements."

For multi-keyword screens, batch separate calls. Comma-separated sources work on get_growth, but each keyword still needs its own request on the amazon source.

Limits and error cases

A 404 not_found response means no series matched the keyword on Amazon. Try a broader category phrase ("collagen powder" instead of a branded ASIN title). Amazon search trends track consumer queries, not individual SKU IDs.

Rate limits follow the account tier documented at https://www.trendsmcp.ai/pricing. Growth calls count the same as trend series calls: one per source plus keyword.