Google Trends Scraper: 200 Keywords on One Scale

What does this Google Trends scraper do?

It compares any number of keywords, from 5 to 200, from Google Trends on one consistent scale, with an error bar for every keyword. It's a Google Trends API alternative without the 5-keyword limit, and small keywords come back as real numbers instead of 0.

In short: up to 200 keywords on one scale in one run, an error bar on every keyword, glitches in Google's data flagged, and keywords with no data are free.

Google Trends compares at most 5 keywords at a time, and it rescales every comparison so its own top keyword is 100. That causes two problems:

1. You can't rank more than 5 keywords. Two separate comparisons use two different scales, so their numbers can't be put side by side.
2. Small keywords disappear. Next to a big keyword, a small one gets rounded down to 0 for the whole year. Its data isn't just imprecise, it's gone.

This Actor fixes both. Send it 5, 50 or 200 keywords and every one comes back on one consistent scale, including the small ones, each with an error bar that tells you how precise it is.

Example: 12 drinks, US, past 12 months

Put "coffee" and "dandelion coffee" into Google Trends directly and dandelion coffee reads 0 every week. Here is the same set, in one run, on one scale:

Rank	Keyword	Avg interest	Share of set	12-mo growth	±
1	coffee	77.846	56.39%	+10.8%	0.10%
2	tea	37.135	26.90%	−2.0%	0.14%
3	espresso	9.615	6.97%	−8.6%	0.42%
4	matcha	6.247	4.53%	+32.8%	0.44%
6	cold brew	2.059	1.49%	+83.3%	0.52%
8	yerba mate	0.500	0.36%	+9.5%	0.75%
11	chicory coffee	0.069	0.05%	−10.2%	1.19%
12	dandelion coffee	0.017	0.01%	0.0%	1.02%

Dandelion coffee is about 1/4,600th of coffee, measured to within about 1%. The run used 7 requests and took 3 seconds.

Why use it?

Error bars you can trust

Every keyword gets an error bar (uncertaintyPct), and it's measured on your run, not guessed:

• Keywords are compared in overlapping groups, and each group shares one keyword with the next. The shared keyword's two copies should match. Any difference beyond what rounding explains is Google's sampling noise, measured directly.
• It was tested against a simulated Google with known true values across thousands of keywords. About 68% of true values land within ±1σ and 95% within ±2σ, which is what honest error bars should do.
• A keyword Google has no usable data for is reported as below_threshold. It is never filled in with a guess.

Built-in check for glitches in Google's data

Sometimes Google Trends shifts across the board. From about late March to late June 2026, unrelated US keywords like shower curtain, doormat and dish rack all roughly doubled together and then fell back. In a 12-month US run of 7 keywords, 5 of them "peaked" inside that stretch, coffee and tea included. Taken at face value, those peak dates are wrong.

Every run checks for this automatically, for free:

• One extra request compares a fixed basket of everyday, unrelated keywords. When they all move together, it's Google's data moving, not your keywords.
• warning tells you when a keyword's peak or growth falls in such a stretch, and growthVsBaselinePct gives the growth with the drift taken out.

Tested every day

An automated check runs this Actor every morning on a real query and checks the data itself (rows, filled fields, error bars), not just that the run finished.

Use cases

• SEO and content: rank a long keyword list by real interest, including the long tail
• Product and market research: compare brands, products or competitors beyond Google's 5-keyword limit
• Trend spotting: find fast-growing keywords early
• Data science: a clean, consistently scaled time series for modelling

What data can you get from Google Trends?

Field	What it is
rank	Position in your set by average interest
averageInterest	Average on the common scale (the single highest point across all your keywords = 100)
shareOfSetPct	Share of all the interest in your set
growthPct	Last quarter of the period vs the first quarter
growthVsBaselinePct	The same growth with Google's market-wide drift taken out
peakInterest, peakDate	The highest point and its date
latestInterest	The most recent complete point
uncertaintyPct	The ± on the keyword's overall level
warning	Set when the peak or growth falls in a market-wide jump in Google's data
status	ok, or below_threshold if Google has too little data
series	Every data point on the common scale, each with marketWideJump

How to scrape Google Trends for many keywords at once

1. Open the Actor and click Try for free.
2. Paste your keywords, one per line (5 to 200).
3. Pick a location and time range (the default is worldwide, past 12 months).
4. Click Start. Most runs take a few seconds.
5. Download the results as JSON, CSV or Excel, or read them through the API.

How much does it cost to scrape Google Trends?

$0.005 per keyword that comes back on the common scale, so 100 keywords cost $0.50. Keywords with no usable data are free, and the glitch check is free. There's also a tiny start fee of $0.00005 per run.

Apify's free plan includes $5 of monthly credit, which covers about 1,000 keywords a month. The run checks your spending limit before it starts, so it never stops halfway through.

Input

See the Input tab for every option. The main ones:

• Keywords: 5 to 200, one per line
• Location: a country or region code (e.g. US, GB, US-CA), or empty for worldwide
• Time range: past 7 days, 30 days, 90 days, 12 months or 5 years, 2004 to now, or a custom date range
• Search type: web, images, news, shopping or YouTube search
• Category: a Google Trends category ID, to narrow ambiguous keywords
• Output format: one row per keyword (with the full series inside), or one row per keyword per date

Output

You can download the dataset as JSON, CSV, Excel, XML or HTML. One row per keyword looks like this (series shortened):

[
  {
    "rank": 1,
    "keyword": "coffee",
    "status": "ok",
    "averageInterest": 54.87,
    "shareOfSetPct": 66.03,
    "peakInterest": 100,
    "peakDate": "2026-07-19",
    "latestInterest": 48,
    "growthPct": -10,
    "growthVsBaselinePct": -2.8,
    "warning": null,
    "uncertaintyPct": 0.06,
    "series": [
      { "date": "2026-07-17", "value": 56, "partial": false, "marketWideJump": false },
      { "date": "2026-07-18", "value": 69, "partial": false, "marketWideJump": false }
    ]
  },
  {
    "rank": 2,
    "keyword": "tea",
    "status": "ok",
    "averageInterest": 23.957,
    "shareOfSetPct": 28.83,
    "growthPct": -11,
    "growthVsBaselinePct": -3.9,
    "uncertaintyPct": 0.13
  }
]

Choose One row per keyword per date to get a flat table (keyword, date, value, ...) that drops straight into Sheets, Excel, Looker Studio or pandas. The OUTPUT record lists the dates of every market-wide jump found (marketWideCheck).

Tips

• How it works: keywords are compared in overlapping groups of similar popularity, so a small keyword is never rounded away by a much bigger one. Each group shares one keyword with the next, and that shared keyword carries the scale across, like a relay race handing over a baton.
• uncertaintyPct is the precision of a keyword's overall level. Individual points of very small keywords are coarser, because Google reports whole numbers.
• When every keyword fits in one group (5 or fewer, of similar size), there are no links to measure sampling noise from, so only rounding error is reported.
• The glitch check uses English control keywords. Outside English-speaking regions they can have too little data, and the check is skipped (checked: false).

FAQ

Why not just compare in groups of 5 myself? Each group is scaled to its own top keyword, so the numbers from different groups can't be compared, and small keywords round to 0 next to big ones.

Is this Google's search volume? No. Google Trends measures relative search interest, and this Actor keeps that meaning while putting everything on one scale.

How many keywords can I send? Up to 200 per run. It uses about one request to Google for every two keywords, plus one for the glitch check.

Can I use it from code or an AI assistant? Yes. Use the Apify API or the Python and JavaScript clients (see the API tab), connect it to Make, Zapier or Google Sheets, or schedule it. AI assistants such as Claude and Cursor can call it through Apify's MCP server: https://mcp.apify.com?tools=honestdata/google-trends-one-scale.

Is it legal to scrape Google Trends? Google Trends data is aggregated and anonymous, and this Actor collects no personal data. As with any data, check that your use fits the rules that apply to you.

Found a problem? Open an issue in the Issues tab. We read every report.

Found it useful? A short review on this page helps other people find it.

More from Honest Data

• AliExpress Product Search Scraper (US): search results with the real regular price, ads and promos flagged.
• YouTube Transcript Scraper: transcripts in bulk at $2 per 1,000, and misses are free.
• Product Demand Radar: rank product ideas by Google demand and AliExpress supply.