How to check if 1,000 keywords are growing on Google Trends in one call

Google Trends is the best free signal for "is demand going up or down?". It is also painful to use at scale:

• the website compares at most 5 keywords at a time;
• every chart is scaled to its own peak (0–100), so you can't read growth off it directly;
• there is no official API, and scripts get rate-limited fast.

If you have a list of 200 product ideas, 500 SEO keywords or a spreadsheet of niches, you usually end up clicking through charts one by one and eyeballing them.

I wanted one row per keyword with a number that answers the question. So I built a small tool for it. This post explains the method, so you can reuse it even without my tool.

The method: compare the last 12 months with the 12 before

Raw Trends values are relative, but ratios inside the same series are meaningful. So for each keyword:

1. Pull 5 years of weekly data for the country you care about.
2. Take the last 52 complete weeks and the 52 weeks before them.
3. Compare the averages: that's your year-over-year growth.

Comparing full years cancels out seasonality. "Sunscreen" in July vs January tells you nothing; this July-to-June year vs the previous one tells you whether the market is actually bigger.

Two extra signals help:

• 3-month momentum: the last 13 weeks vs the 13 weeks before. Is the growth still accelerating, or already cooling?
• Current vs peak: a keyword at 100% of its peak is a different story from one at 30%.

Then put everything into plain labels that anyone on the team can read:

Real example (US, November 2026)

How to read it:

• "home battery backup" more than doubled over a year, but momentum is down 46% over the last quarter. The demand is real, but the spike is behind it. That's useful before ordering stock or writing 20 articles on the topic.
• "air fryer" is a mature market: slightly up, flat overall. Big volume, no growth story.
• The growth score (0–100, 50 = flat) combines year-over-year growth and recent momentum into a single sortable number. Sort your list by it and the interesting keywords float to the top.

Bonus: why did it spike?

A number tells you that a keyword jumped, not why. So the tool can also look at news headlines from the weeks around each keyword's biggest spike and extract the phrases that keep coming up:

I tested it on 50 well-known spikes in English, French and Arabic: it found the right cause in 43 of 50. Not perfect, but enough to separate a one-off news event from a lasting trend without opening 30 tabs.

Doing it in one call

The tool runs on Apify as Google Trends Growth Checker. Input:

{ "keywords": ["home battery backup", "portable power station", "heat pump rebate"], "countries": ["US", "GB"], "includeRelatedQueries": true, "explainWithNews": false } 

Up to 1,000 keywords per run, each one checked in every country you list. From Python:

from apify_client import ApifyClient client = ApifyClient("YOUR_APIFY_TOKEN") run = client.actor("easydata-labs/keyword-growth-checker").call(run_input={ "keywords": ["home battery backup", "portable power station"], "countries": ["US"], }) for row in client.dataset(run["defaultDatasetId"]).iterate_items(): print(row["keyword"], row["trend"], row["growthScore"], row["yoyChangePct"]) 

Each row also includes the 10 fastest-rising related searches, which is often where the next keyword idea comes from, plus a lowVolumeWarning flag when Google has too little data for the number to be trusted.

It also works from n8n, Make, or an AI agent through Apify's MCP server: the agent gets a clean, labeled row instead of a raw chart to interpret.

Limits worth knowing

• Google Trends measures relative interest, not search volume. Use this to see whether demand is growing; use a search-volume tool to see how big it is.
• Very small keywords are noisy. Respect the low-volume flag.
• Growth is computed on complete weeks, so the current partial week is ignored on purpose.

Pricing

Pay per keyword: $12 per 1,000 keywords checked, and failed checks are not charged. The news explanation is optional and charged separately ($15 per 1,000), only when an explanation is actually found.

Disclosure: I built this tool. If you try it and something is off, or you need a field it doesn't have, leave a comment. I read them all.