Growth engineering notebooks that treat trends as test inputs

Why do experiment docs arrive without shared demand context?

Growth engineers often inherit a hypothesis sentence and a spreadsheet of internal metrics, while external curiosity lives in separate tabs nobody copies into the ticket. Trends MCP lets a notebook or agent fetch google search, tiktok, and amazon series with the same JSON fields, which makes preflight reviews faster and calmer.

What does a typical pre launch pull look like?

An engineer posts three JSON bodies to the API: Get Growth on google search for the campaign theme with 14D, 30D, and 3M presets, Get Trends on tiktok for the hashtag the creative team wants to own, and Get Trends on amazon for the hero SKU phrase. If video is part of the test, add youtube for the same phrase. The response objects include dated points the team can plot quickly in Python or paste into a review doc.

How do Zapier and Cursor fit real teams?

Operations heavy groups can follow https://www.trendsmcp.ai/zapier-trend-workflows to schedule pulls when spend switches on. Engineers who draft in the IDE can use the Cursor block at https://www.trendsmcp.ai/mcp-server-for-cursor so the assistant never improvises volumes from training data.

What should experiment owners still do manually?

Trends MCP cannot read private funnel tables, incrementality results, or finance approvals. It belongs in the chapter of the doc that explains why the outside world might care, while warehouse data still owns conversion truth. Say both parts aloud in experiment retros so stakeholders do not confuse attention with revenue.