Google Trends API – Interest, Regions & Related

Get Google Trends data at scale – interest over time, interest by region, related queries and related topics – for any number of search terms, with comparisons Google's own UI cannot do.

What makes it different

• Compare unlimited terms on one scale. Google compares at most 5 terms; this Actor chains batches through an anchor term so 20 or 200 terms stay comparable.
• Same term across countries or years on one scale – e.g. bitcoin in US vs DE vs PL, or Black Friday 2024 vs 2025.
• Daily data for 5+ years (or hourly data for weeks): overlapping windows are stitched onto one 0–100 scale.
• Related topics that actually return data – 20 of 20 terms in our last measurement (2026-10-20), where other Google Trends Actors we measured returned empty lists – with topic ids you can reuse as search terms.
• Topics as input – use /m/0k8z (Apple – Technology company) instead of an ambiguous word; mode suggest finds the ids.
• All 1,426 categories, 3,600+ locations (countries, states, US metro areas, cities) and all search types: web, images, news, shopping, YouTube.
• What people search for without a keyword – top and rising queries for a country and category.
• Google's methodology notes (dataNotes) – the "Note" markers that explain jumps caused by changes in Google's data collection.
• Honest results: every term gets a status (ok, no_data, blocked, error). You pay only for terms with data; the run stops on time (maxRunSecs) with partial results instead of hanging.

Who uses it

• SEO and content teams – which keywords are growing, where, and what people search for next (rising queries and topics).
• Market research and product teams – compare brands, products or categories across countries and years on one scale.
• E-commerce – seasonality: when demand for a product starts, peaks and fades, by state or city.
• Finance and data science – long daily series of search interest as a signal for models and dashboards.
• AI agents and automations – a documented input schema, one row per term with a status, and a run summary for decisions.

Quick start

{
    "searchTerms": ["bitcoin", "ethereum"],
    "geo": "US",
    "timeRange": "past-12-months"
}

More examples:

Goal	Input
10 brands on one scale	{"searchTerms": ["nike","adidas","puma","asics","reebok","new balance","under armour","fila","vans","converse"], "comparison": "together"}
One term in three countries	{"searchTerms": ["ozempic"], "comparison": "acrossGeos", "geos": ["US","GB","DE"]}
Year over year	{"searchTerms": ["black friday"], "geo": "US", "comparison": "acrossTimes", "timeRanges": ["2024-10-01 2024-12-31", "2025-10-01 2025-12-31"]}
Daily data for 5 years	{"searchTerms": ["chatgpt"], "timeRange": "past-5-years", "granularity": "daily"}
Trending searches in a category	{"geo": "US", "category": 71} (no search terms)
YouTube searches by US metro area	{"searchTerms": ["minecraft"], "geo": "US", "property": "youtube", "regionResolution": "DMA"}
States, cities and metros at once	{"searchTerms": ["coffee"], "geo": "US", "regionResolution": "REGION", "extraRegionResolutions": ["CITY", "DMA"]}
Safe limit for long lists	{"spreadsheetUrl": "https://docs.google.com/spreadsheets/d/<id>/edit", "maxTerms": 500}
Re-run a link from the browser	{"startUrls": [{"url": "https://trends.google.com/trends/explore?q=pizza&geo=IT"}]}
Terms from a spreadsheet	{"spreadsheetUrl": "https://docs.google.com/spreadsheets/d/<id>/edit"} (shared as "anyone with the link")
Find topic ids	{"mode": "suggest", "searchTerms": ["apple", "jaguar"]}
List categories / locations	{"mode": "reference", "referenceType": "geos"}

Search operators work as in Google Trends: "exact phrase", tennis + squash (either), jaguar -car (exclude).

Output

One row per term (layout nested). A real result (run of 2026-10-16, arrays shortened with "…"; full run: 53 weekly points, 51 regions (states + DC), 25 + 25 related queries, 25 + 18 related topics):

{
    "term": "solar panels",
    "termType": "query",
    "geo": "US",
    "timeRange": "today 12-m",
    "category": 0,
    "property": "web",
    "comparison": "separate",
    "comparedWith": [],
    "status": "ok",
    "average": 35.34,
    "peak": 100,
    "peakDate": "2026-08-09",
    "latest": 27,
    "changePct": 15.9,
    "direction": "rising",
    "timeline": [
        { "date": "2025-10-12", "value": 27, "hasData": true, "isPartial": false },
        { "date": "2025-10-19", "value": 23, "hasData": true, "isPartial": false },
        "…",
        { "date": "2026-10-11", "value": 24, "hasData": true, "isPartial": true }
    ],
    "dataNotes": [],
    "regions": [
        { "geoCode": "US-WY", "geoName": "Wyoming", "value": 100 },
        { "geoCode": "US-KS", "geoName": "Kansas", "value": 82 },
        "…"
    ],
    "relatedQueries": {
        "top": [
            { "title": "solar panels cost", "value": 100, "formattedValue": "100" },
            { "title": "solar panel", "value": 60, "formattedValue": "60" },
            "…"
        ],
        "rising": [
            {
                "title": "flooded solar panels arc raiders",
                "value": 7800,
                "formattedValue": "Breakout",
                "isBreakout": true
            },
            "…"
        ]
    },
    "relatedTopics": {
        "top": [
            { "title": "Solar energy", "mid": "/m/06xbd", "topicType": "Topic", "value": 100 },
            { "title": "Solar panel", "mid": "/m/0fnrbc", "topicType": "Topic", "value": 84 },
            "…"
        ],
        "rising": [
            { "title": "Solar panel", "mid": "/m/0fnrbc", "topicType": "Topic", "formattedValue": "Breakout" },
            "…"
        ]
    },
    "warnings": [],
    "trendsUrl": "https://trends.google.com/trends/explore?date=today+12-m&geo=US&q=solar+panels",
    "scrapedAt": "2026-10-16T10:04:05.553Z"
}

Every row has a stable id: the same term in the same comparison gets the same id in every run, so scheduled runs can be deduplicated or upserted. Additional region resolutions come in regionsByResolution.

Layout flat gives one row per data point – timeline dates, regions, related queries and related topics, told apart by rowType – ready for Google Sheets, Excel or BI tools. Each flat row has its own stable id too.

Each run also stores a summary in the OUTPUT record of its key-value store (items per status, related topics blocked, why the run stopped, charged events) – one read for integrations and AI agents.

Metrics: average and peak over the period, latest complete value, changePct = average of the last quarter of the period vs. the quarter before it, direction = rising / falling (±10 %) or stable. Values are relative (0–100) as in Google Trends: each point is the share of all searches in that place and time, scaled to the highest point of the comparison – not a number of searches. In acrossGeos a higher value means a larger share of searches in that country, not more searches. Stitched and anchored series keep two decimals.

Integrations

• API: POST https://api.apify.com/v2/acts/datarelay~google-trends-api/run-sync-get-dataset-items with the input as JSON body.
• AI agents (MCP): add the Actor in the Apify MCP server – every input field is documented for the model.
• n8n / Make / Zapier: use the Apify integration, action "Run Actor", then read the dataset items.
• Schedules: run daily/weekly with the same input and get only fresh data.

Pricing – what a job costs

You pay only for results with data; platform usage is included (no compute or proxy costs on top).

Event	Price (FREE / BRONZE)	When
run-start	$0.02	once per run, together with the first paid result
term	$1.20 / 1,000	term with data: timeline, regions, related queries
related-topics	$10 / 1,000	term whose related topics came back (not charged when they stay blocked)
term-window	$1 / 1,000	each extra window of daily / hourly data over long ranges
flat-row	$0.01 / 1,000	each row of the flat layout (a term with all sections ≈ 200 rows ≈ $0.002)
suggestion	$0.50 / 1,000	suggest lookup that found topics

SILVER and GOLD plans get lower term, term-window and suggestion prices (−3 to −8 %). reference mode is free.

Typical job	Cost
1 term, 12 months, all sections incl. related topics	$0.031
1 term without related topics	$0.021
100 terms with related topics	$1.14
100 terms without related topics	$0.14
10 terms on one scale (timeline, regions, queries)	$0.032
Daily data for 5 years, 1 term (10 windows)	$0.030

You don't pay for terms without data (no_data), blocked or failed terms, related topics that stayed blocked, or a run that returns no paid result (no start fee either).

Reliability (measured)

• Load test of 2026-10-27 (112 runs, 306 terms, real inputs at 10× our expected traffic): 98.7 % of timelines, 98.6 % of regions and related queries and 96 % of related topics delivered. The rest were terms Google rate-limited on every attempt; they are reported with status: "blocked" (or a section warning) and not charged.
• Related topics come only through Google's embed path, which refuses some IPs. The Actor retries on other IPs, tries once more at the end of the run (at least 30 s later – blocks come in waves) and, if that fails too, delivers the term with the warning relatedTopics: blocked (and does not charge related-topics).
• Speed: a run with 1–5 terms takes ~10 s (median); 20 terms take ~1 minute without related topics and ~2 minutes with them (measured 2026-10-27: 55 s and 135 s, all 20 terms delivered).
• Scheduled control runs check the Actor every hour, with alerts on Google's rate limits.

Limits and good to know

• Values are relative (0–100), not search counts – see Output. Compare terms only within one comparison (together, acrossGeos, acrossTimes).
• Google rate-limits by IP. By default the Actor uses Apify residential proxies: each term runs on its own IP, and when Google blocks it the term continues on a fresh IP from where it stopped (never on the blocked one); the last attempts use your default Apify Proxy. If your account cannot use residential proxies, the run uses your default Apify Proxy (more blocks on large runs).
• When a section stays blocked after all retries, the term is still delivered with a warning such as relatedTopics: blocked or regions: blocked; only a blocked timeline makes the term blocked (not charged). Terms with blocked related topics are delivered at the end of the run, after one more try.
• When Google blocks every term of a run, the run still finishes as succeeded, with a blocked row per term, the status message "No results…" and outcome: "no_results" in the OUTPUT summary – nothing is charged; try again later.
• More than 5 terms on one scale: terms are compared in batches of 5 linked through an anchor term, chosen automatically (the first term, or the most popular one when the first is too rare – a few extra requests). Terms thousands of times less popular than the top term show 0, as they would in Google Trends.
• acrossTimes compares at most 5 periods (Google puts at most 5 on one scale).
• Google computes Trends from a sample: the same request can return slightly different values on different runs.
• Google's rising lists sometimes contain unrelated or spam-like breakout queries (e.g. "home workout routines" for coffee in one of our runs) – Google's data, not an Actor error; filter them if you use them automatically.
• Google returns integers; very small terms may show 0 with hasData: false (too little data).

FAQ

Does it give search volumes? No – Google Trends gives relative interest (0–100), not counts: see Output. For volume ranges of trending searches ("20K+"), use our Google Trending Now Actor.

Why not pytrends or my own script? They run from your IP, and Google blocks it after a few dozen requests. This Actor spreads requests over residential IPs, retries blocked requests on fresh ones, compares more than 5 terms on one scale and returns related topics.

Why do two runs give slightly different values? Google computes Trends from a sample of searches, so the same request can differ by a few points between runs. Trends and shapes stay the same.

What does blocked mean? Google rate-limited every attempt for that term. You are not charged; run it again later (blocks usually pass within minutes to hours).

How do I get the data into Google Sheets or Excel? Use layout flat (one row per data point) and export the dataset as CSV / Excel, or add Apify's Google Sheets integration. Terms can also come from a sheet (spreadsheetUrl).

Can I schedule it? Yes – create a task with your input and a schedule; stable row ids let you deduplicate or upsert results.

Not affiliated with Google

This is an independent Actor by DataRelay, not Google's official Google Trends API (which is in a limited alpha). It reads the same data that the public Google Trends website shows.

Usage statistics

The Actor sends anonymous run statistics to the developer (mode and options, number of items and their statuses, Google request counts, duration). Search terms and your Apify user id are never sent – only keyed hashes, used to count repeated queries and distinct users. No results, account details or personal data are collected.

Changelog

• 0.1 – first version.