Reliable Google Trends API

Get Google Trends data as clean, flat JSON: interest over time, interest by region, related and rising queries, and related topics for single keywords or comparisons of up to 5 keywords, plus Trending now lists by country. Google throttles automated Trends requests hard, and that is why Trends scrapers fail. This actor is built around that throttling: it retries patiently, finishes the run, and never charges for a keyword it could not deliver.

Features

• Keyword batches in one run: any number of searches, each a single keyword or a comparison of up to 5 keywords on one shared 0-100 scale.
• Everything trends.google.com shows for a keyword in one record: the time series, the regional breakdown (countries, states/regions, or cities, depending on the region you pick), top and rising related queries, and related topics when Google provides them.
• Trending now by country: every trending search with its search volume, growth, start time, categories and related searches.
• Flat, predictable JSON for spreadsheets, scripts and AI agents. Empty sections come with a plain-English notes entry saying why they are empty.
• Built for reliability: retries per request, then a second pass at the end of the run for anything still throttled. A keyword that still fails is reported with its error and not charged.
• Web, YouTube, News, Image and Google Shopping search, 25 categories, every preset time range, and custom date ranges.

How it works

1. Put your keywords in Search terms: one search per line. Put up to 5 keywords on one line, separated by commas, to compare them.
2. Choose the region, time range, category and Google property, the same choices as on trends.google.com.
3. The actor asks Google Trends for the same data the trends.google.com page loads, through residential proxies with real-browser TLS fingerprints. When Google throttles a request, it waits and retries on the same connection before switching to a fresh one. In our tests that mattered far more than which proxy we used.
4. Each keyword becomes one dataset record as soon as all of its data has arrived. You are charged only then.

For Trending now, switch Mode to Trending now and list the countries.

Input

Field	Type	Default	Description
mode	enum	keywords	keywords or trendingNow.
searchTerms	array	["coffee"]	One search per line; commas inside a line compare up to 5 keywords.
geo	string	"" (worldwide)	Country code (US, GB, DE), sub-region (US-CA, GB-SCT), or empty for worldwide: the geo= value of a trends.google.com URL.
timeRange	enum	today 12-m	now 1-H, now 4-H, now 1-d, now 7-d, today 1-m, today 3-m, today 12-m, today 5-y, all.
customTimeRange	string		Overrides timeRange: 2024-01-01 2024-12-31, or with hours for ranges under a week: 2026-10-01T00 2026-10-05T23.
category	enum	"" (all)	Google Trends top-level category id, e.g. 71 = Food & Drink.
property	enum	"" (web)	youtube, news, images, or froogle (Google Shopping).
language	string	en-US	Language for region and topic names.
trendingCountries	array	["US"]	Trending now: country codes, one list per country.
trendingHours	enum	24	Trending now: 4, 24, 48 or 168 hours.
trendingMaxItems	integer	0 (all)	Trending now: cap on trends per country.

Example 1, one keyword in the US over the past 12 months (the default run, with geo set):

{
    "searchTerms": ["coffee"],
    "geo": "US"
}

Example 2, a 5-keyword comparison plus a separate keyword, on YouTube, over 5 years:

{
    "searchTerms": ["coffee, tea, matcha, espresso, chai", "cold brew"],
    "geo": "US",
    "timeRange": "today 5-y",
    "property": "youtube"
}

This returns 6 records: five sharing one 0-100 scale, and "cold brew" on its own scale.

Example 3, trending searches in the US and Germany over the last 4 hours:

{
    "mode": "trendingNow",
    "trendingCountries": ["US", "DE"],
    "trendingHours": "4"
}

From code (Python, pip install apify-client):

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("docfoundry/google-trends-api").call(
    run_input={"searchTerms": ["coffee, tea"], "geo": "GB", "timeRange": "today 3-m"})
for record in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(record["keyword"], record["averageInterest"], record["peakDate"])

Output

Keywords mode: one record per keyword. A real record (trimmed: the full series has 53 weekly points and the regions list has all 51 states):

{
    "type": "keyword",
    "keyword": "tea",
    "comparedWith": ["coffee", "matcha", "espresso", "chai"],
    "geo": "US",
    "timeRange": "today 12-m",
    "category": 0,
    "property": "web",
    "status": "ok",
    "averageInterest": 37,
    "peakDate": "2026-04-12",
    "interestOverTime": [
        { "date": "2025-10-19", "value": 33 },
        { "date": "2026-10-18", "value": 33, "isPartial": true }
    ],
    "regionResolution": "REGION",
    "interestByRegionScale": "shareOfComparison",
    "interestByRegion": [
        { "geoCode": "US-SD", "name": "South Dakota", "value": 42 },
        { "geoCode": "US-MS", "name": "Mississippi", "value": 33 }
    ],
    "relatedQueriesTop": [
        { "query": "green tea", "value": 100 }
    ],
    "relatedQueriesRising": [
        { "query": "how to make cold brew tea", "increasePercent": 2050, "breakout": false, "formattedValue": "+2,050%" }
    ],
    "relatedTopicsTop": null,
    "relatedTopicsRising": null,
    "notes": [
        "Related topics are null because Google Trends does not provide them for comparisons. Put the keyword on its own line to request them."
    ],
    "trendsUrl": "https://trends.google.com/trends/explore?q=coffee%2Ctea%2Cmatcha%2Cespresso%2Cchai&date=today%2012-m&geo=US",
    "scrapedAt": "2026-10-19T22:20:33Z"
}

Field	Meaning
interestOverTime	{date, value} points. date is a day (2025-10-19, the start of a week or month for longer ranges) or, for ranges under a week, a UTC time (2026-10-12T22:00:00Z). isPartial: true marks the current, still-incomplete period; hasData: false marks a point Google has no data for.
averageInterest	Google's own average for comparisons; for a single keyword, the mean of the series.
peakDate	The date of the highest point.
interestByRegion	Regions sorted by value. regionResolution says what they are: COUNTRY (worldwide), REGION (states/provinces), DMA (US metros) or CITY (with lat/lng instead of a code). Regions without data are left out.
relatedQueriesTop	Most frequent related searches, scored 0-100 against the top one.
relatedQueriesRising	Fastest-growing related searches: increasePercent is the growth over the previous period; breakout: true is Google's "Breakout" label (growth over 5000%).
relatedTopicsTop / relatedTopicsRising	{topic, type, mid, value}; mid is the Google Knowledge Graph id. null in comparisons (see Limitations).
notes	Why a section is empty. Always read it before treating an empty list as zero interest.
status / error	ok, or failed with the error. Failed records are never charged.
trendsUrl	The same view on trends.google.com, to check any number by eye.

Reading Google's 0-100 numbers

Google Trends never reports search counts for keywords. Every number is relative:

• Over time: 100 is the highest point in the chosen range and region; 50 means half as popular as that peak. In a comparison, all keywords share one scale: 100 is the single highest point of any of them, so you can compare keywords directly. The same keyword gets different numbers in a different comparison, region or time range.
• By region, single keyword (interestByRegionScale: "relativeToTopRegion"): 100 is the region where the keyword makes up the largest share of all searches. That is search share, not volume, so a small state can top the list.
• By region, comparison ("shareOfComparison"): each keyword's share of the compared searches in that region. The values of all keywords in a region add up to 100.
• 0 means too little data, not necessarily zero searches.

Trending now mode: one record per country, with the trends in a list (trimmed to one trend):

{
    "type": "trendingNow",
    "country": "US",
    "windowHours": 24,
    "status": "ok",
    "trendCount": 429,
    "trends": [
        {
            "rank": 2,
            "query": "davante adams",
            "searchVolume": 20000,
            "increasePercent": 1000,
            "startedAt": "2026-10-19T20:50:00Z",
            "endedAt": null,
            "isActive": true,
            "categories": ["Sports"],
            "relatedQueries": ["poona ford", "ruptured hamstring", "kobie turner"],
            "trendsUrl": "https://trends.google.com/trends/explore?q=davante%20adams&geo=US&date=now%207-d"
        }
    ],
    "notes": [],
    "scrapedAt": "2026-10-19T22:20:33Z"
}

searchVolume is Google's rounded lower bound ("20K+" searches); increasePercent is the growth Google shows next to it; endedAt is set once a trend is no longer active.

The dataset has two table views, Keywords and Trending now (one row per trend), and exports to JSON, CSV and Excel. For a flat file with one row per trend, add unwind=trends to the export URL, as the run's Trending now (CSV) output link does.

Pricing

Pay per event, no subscription. Prices below are for Apify's free plan; paid Apify plans pay less per result: $0.0027 (Starter), $0.0024 (Scale) and $0.002 (Business). Platform usage, including the residential proxy, is included. Put many keywords in one run: the run-start fee is charged once per run.

Event	Price	When it is charged
Run start	$0.01	Once per run, when it starts (Apify's standard apify-actor-start event).
Keyword result	$0.003	Once per keyword, after all its data arrived. 5 keywords in one comparison = 5 results.
Trending list	$0.003	Once per country list, however many trends it contains.

Monthly use	Cost
250 keywords in 25 runs	$1.00
1,000 keywords in 100 runs	$4.00
30 daily trending lists (30 runs)	$0.39

Failed keywords are never charged. Low-volume keywords that Google answers with empty data are charged as normal results, because that answer is real data; their notes say so. Set a maximum charge in the run options and the actor stops fetching before it would go over: a comparison that does not fit under the limit is skipped, not half-done.

Limitations

• Related topics are mostly empty right now. As of October 2026 Google returns an empty related-topics list for most keywords, popular ones included, and sends no topics at all for comparisons (null). The fields fill in whenever Google sends topics. Related queries are not affected.
• Relative numbers only. Google Trends does not publish absolute search volumes for keywords (see "Reading Google's 0-100 numbers"). Trending-now volumes are Google's rounded buckets.
• At most 5 keywords per comparison, as on trends.google.com. Comparing more needs a shared anchor keyword across several lines.
• Speed. A keyword takes about 2-5 seconds. When Google throttles, the waits between retries stretch that to a minute or more, and a heavily throttled task gets one more try at the end of the run. 100 comparisons usually finish in a few minutes; the default run timeout is 1 hour.
• Not 100%. In our 200-task tests on 2026-10-19, 195-200 of 200 tasks succeeded on the first pass, depending on how hard Google was throttling at the time. The end-of-run retry pass was added for exactly those cases. Anything that still fails is listed with status: "failed", logged, and not charged.
• Per-keyword maps inside a comparison (trends.google.com's small map under each keyword) are not fetched; the compared regional breakdown is.
• Region codes are Google's: ISO country codes, ISO sub-regions like US-CA, and Nielsen metro codes like US-NY-501.

Data privacy

This actor reads only public, aggregated Google Trends data. It needs no Google account and no cookies of yours, and it handles no personal data. Your search terms are sent to Google Trends only to fetch their data. Results live only in your own run's dataset, under your Apify account's retention settings.

Screenshots

Compare up to 5 keywords per line, as many lines as you like:

One row per keyword, on the same 0-100 scale, with series, regions and related queries:

Trending now, one row per trend with volume, growth and related searches:

Support

Open an issue on the actor's Issues tab, ideally with the run link. A keyword that fails repeatedly, or a field that looks wrong next to trends.google.com, is exactly what we want to hear about. Typical response: 1 business day.