Google Trends Scraper — bulk, 5+ terms on one scale

Google Trends data for many search terms at once: interest over time, interest by region (countries, states, US metro areas, cities) and related queries. It reads Google's own data endpoints directly, without a browser, so it is fast and cheap to run. Terms Google has no data for, and terms that could not be read, are delivered free with the reason. You can also put any number of terms on one 0-100 scale (Google's chart stops at 5), and get daily data for ranges up to 5 years (Google switches to weekly after 270 days).

Versión en español más abajo.

Quick facts

• What it does: Google Trends for many terms at once, without a browser: interest over time, interest by region, related queries (and topics); any number of terms on one 0-100 scale; daily data for ranges up to 5 years.
• Use it for: keyword and market research in bulk; comparing many terms on one scale; long daily series; regional interest (countries, regions, US metros, cities).
• Not for: absolute search volumes (Google's numbers are relative, 0-100); Trending Now; hourly stitching or daily data past 5 years (not in this version).
• Input: searchTerms (required; up to 5,000 term × location pairs per run); geos (Trends codes; empty = worldwide); timeRange (default today 12-m, or YYYY-MM-DD YYYY-MM-DD); optional compareTogether + anchorTerm, dailyForLongRanges, section switches, regionResolution, category, searchType.
• Output: one dataset row per term and location (rows without data come free, with status and the reason), and one SUMMARY record. Every field with its type is under Output reference below.
• Charged events (pay per event): keyword USD 0.003 per term with data (USD 3.00 per 1,000); keyword-daily USD 0.01 per term when dailyForLongRanges is on and the range is longer than 270 days (USD 10.00 per 1,000); a term is charged when its interest over time came back (or any section, if the timeline is off); plus Apify's standard start event apify-actor-start.
• Never charged: terms without data in Google (noData), terms that could not be read (failed), the SUMMARY record; there is no proxy or compute charge on top.
• Limits: up to 5,000 term × location pairs per run; stops at your maximum charge per run (SUMMARY stoppedBy); Google samples searches and adds small random noise, so another day can differ a little.
• Run it from: Apify Console, the Apify API or an AI agent through the Apify MCP server (Actor pampadata/google-trends-scraper), with schedules and integrations.

Minimal input

A small input that works, taken from this Actor's own tests; every other field is optional:

{
  "searchTerms": [
    "bitcoin",
    "ethereum",
    "solana"
  ],
  "geos": [
    "US"
  ]
}

Why this one

• Built for volume. No browser: 175 terms (timeline, regions and related queries for each) in 5 minutes in one test, every one complete. When Google slows the run down, the Actor moves to fresh residential IPs by itself. You pay nothing extra for that: there is no proxy or compute charge on top of the price.
• You pay only for data. A term is charged when its interest over time came back (or, if you did not ask for the timeline, when any section did). A term with no data in Google, or one that could not be read after several tries, is still delivered so you know about it, with status noData or failed and the reason. It is not charged.
• A status for every section. Each row says, for the timeline, regions, related queries and related topics, whether it is ok, empty (Google answered with nothing), failed or notOffered. Empty lists never pass for data. The SUMMARY record lists every term that was not complete, with the reason.
• More than 5 terms on one scale. Turn on compareTogether. Up to 5 terms come from one Google chart. With more, every batch of 5 includes an anchor term, and its values tie the batches to one scale. Test on 2026-10-09 with 12 foods: the chained values of empanada and falafel were within 0.6 % of what Google shows when those terms are charted together. In Google's own 5-term chart with pizza, falafel is all 0. Values on the shared scale keep decimals.
• Daily data for long ranges. Turn on dailyForLongRanges to get one point per day for up to 5 years. Google only gives daily points for 270 days or less. The Actor reads overlapping daily windows and fits them to Google's own weekly series for the whole range. Every row includes the check (stitchCheck): the average gap to Google's weekly values on the 0-100 scale. In tests it was 1.0 point over 3 years and 2.6 over 5 years.
• Any location code. It takes any code Google Trends takes: countries (US, GB, KR…), regions (US-CA, GB-ENG, BR-SP), US metros (US-NY-501), or worldwide. There is no fixed list with broken entries.

Input

Field	What it does
searchTerms	Words or phrases (up to 5,000 term × location pairs per run). Google's syntax works ("a + b"; topic ids like /m/0vpj4_b).
geos	Location codes; empty = worldwide. Each term is read for each location.
timeRange	now 1-H, now 4-H, now 1-d, now 7-d, today 1-m, today 3-m, today 12-m (default), today 5-y, all, or your dates: 2024-01-01 2025-06-30.
compareTogether, anchorTerm	All terms on one scale (up to 200 terms); the anchor ties batches of 5 (default: the first term). Best anchor: a term of middle popularity among yours.
dailyForLongRanges	Daily points for ranges up to 5 years (charged as keyword-daily).
includeTimeline, includeByRegion, includeRelatedQueries, includeRelatedTopics	Sections to read. Related topics are off by default: Google currently offers them empty to sessions that are not signed in.
regionResolution	auto, COUNTRY (worldwide only), REGION, DMA (US metros), CITY. includeLowVolumeRegions adds places with too little data.
category, searchType	Google Trends category number (0 = all); web, images, news, YouTube or Google Shopping search.

Output (one row per term and location)

term, geo, timeRange, category, searchType, status, sections; resolution; interestOverTime (date (UTC), value, isPartial on the last point while that period is still running), averageInterest, peakDate; interestByRegion (geoCode, geoName, value; lat/lng for cities); relatedQueries and relatedTopics (top and rising, with breakout: true when Google shows "Breakout"). When comparing, the row also has comparedWith, sharedScale, anchorTerm and lowPrecision. lowPrecision means Google's whole numbers carry a rounding of several %. With daily data, the row has stitchCheck.

{
  "term": "bitcoin", "geo": "US", "timeRange": "today 12-m", "status": "ok",
  "sections": {"timeline": "ok", "byRegion": "ok", "relatedQueries": "ok"},
  "resolution": "WEEK", "averageInterest": 35.56, "peakDate": "2026-02-01",
  "interestOverTime": [{"date": "2025-10-05", "value": 39}, {"date": "2026-10-04", "value": 17, "isPartial": true}],
  "interestByRegion": [{"geoCode": "US-WY", "geoName": "Wyoming", "value": 100}],
  "relatedQueries": {"top": [{"query": "bitcoin price", "value": 100, "formattedValue": "100"}], "rising": []}
}

Price

You pay per term with data: USD 3.00 per 1,000 terms (USD 0.003 each; one term in one location is one row), or USD 10.00 per 1,000 terms with daily data for long ranges (dailyForLongRanges on and a range longer than 270 days; USD 0.01 each). A term is charged when its interest over time comes back (or, if you turned the timeline off, when any section comes back). No charge per run beyond Apify's standard start event, and nothing for the SUMMARY record, terms without data in Google, or terms that could not be read. Set a maximum charge per run in Apify and the Actor stops there.

How long it takes

3 terms in about 5 seconds; 175 terms in about 5 minutes. Daily data for 5 years: about 6 seconds per term.

Good to know

• Google's numbers are relative, 0-100 within each chart: 100 is the highest point of that search, place and time. Rows read separately are each on their own scale. Use compareTogether to compare terms.
• Region values are always on the term's own scale, also when comparing. Google draws each term's map on its own.
• Related queries are what Google limits hardest. In a 60-term test, 3 terms came back without them after 8 tries. Those rows are partial, and the SUMMARY says which section is missing. They are charged, because their timeline came back.
• The last point can be partial: the week or day is still running. It is marked isPartial and left out of averageInterest.
• Times are UTC.

Reliability

What was measured before publishing (2026-10-09), and what was not:

• Runs: every test run finished. 175 of 175 terms came back with their timeline, regions and related queries; in a later 60-term run, 3 terms came back without related queries after 8 tries (partial, timeline and regions present). A term that cannot be read is delivered with status: failed and the reason, and is not charged.
• Blocking: Google refuses one IP after about 25 terms in a row. The Actor then moves to fresh residential IPs by itself: 200 of 200 terms came back that way in a test. If Google also refuses those, the terms come back failed (free), not as a broken run.
• Same question, same numbers: asked twice the same day, and with shifted windows, Google gave identical weekly values (0 difference in 261 weeks). This was not measured across days. Google says Trends works on a sample of searches and adds "small and random fluctuations" to protect privacy, most visible for low-interest terms (Google Trends Help), so a run on another day can differ a little.
• Whole numbers: Google gives 0-100 as whole numbers. For a term that peaks under 10 that rounding is several %; such rows are marked lowPrecision.
• Not measured yet: the success rate on the Store over weeks of real use (published 2026-10-09). For comparison, the most used Google Trends Actor fails or times out in about 1 run of 4 (its public run statistics, read 2026-10-09).

Not affiliated

Not affiliated with, endorsed by or sponsored by Google. "Google Trends" names the public data source.

Output reference

Dataset: one row per term

Field	Type	What it is
term	string	The search term.
geo	string	Location code, or "worldwide".
timeRange	string	Time range sent to Google.
category	integer	Google Trends category (0 = all).
searchType	string	web, images, news, youtube or shopping.
status	string	ok, partial (some section failed), noData or failed (noData/failed rows are not charged).
sections	object	Status per section: timeline, byRegion, relatedQueries, relatedTopics = ok, empty, failed or notOffered.
resolution	string or null	Point spacing (e.g. WEEK, DAY; "(stitched from daily windows)" with dailyForLongRanges).
interestOverTime	array of object	Points: date (UTC), value (0-100), isPartial (true on a still-running last point).
averageInterest	number or null	Average of the points (partial point left out).
peakDate	string or null	Date of the highest point.
interestByRegion	array of object	geoCode, geoName, value (0-100 on the term's own scale); lat/lng for cities.
regionResolution	string	Region level used (auto, country, region, dma, city).
relatedQueries	object	top and rising lists: query, value, formattedValue; on rising, breakout true when Google shows "Breakout" (over 5,000 %).
relatedTopics	object	Only when asked: top and rising topics (topic, topicType, topicId, value, formattedValue); Google often offers none to signed-out sessions.
comparedWith	array of string	Only with compareTogether: the terms on the same scale.
sharedScale	boolean	Only with compareTogether: true when the values are on the shared scale.
anchorTerm	string	Only with compareTogether and an anchor: the anchor term.
lowPrecision	boolean	Only when comparing: true when Google's whole numbers carry a rounding of several %.
stitchCheck	object	Only with dailyForLongRanges: the check against Google's weekly series (average gap on the 0-100 scale).
note	string	Explanation when a row has no usable data.
error	string	Only on failed rows: why the term could not be read.
errors	array of string	Section errors on partial rows.

SUMMARY record (run's key-value store, never charged)

Field	Type	What it is
terms	integer	Number of terms asked.
locations	array of string	Locations asked ("worldwide" when empty).
timeRange	string	Time range used.
dailyStitched, compareTogether	boolean	Modes used.
anchorTerm	string or null	Anchor used when comparing.
rows	object	Counts: ok, partial, noData, failed, notRun, deliveredWithData, charged.
notDeliveredOrIncomplete	array of object	Every term that was not complete: term, geo, status, why.
warnings	array of string	Warnings.
requests	object	Request counters of the run (requests, retries, refusals or blocks, failures; whether the run moved to residential IPs).
finished	boolean	true when the run ended normally.
stoppedBy	string or null	"your maximum charge per run was reached" when it stopped there.

More Actors by this developer

• Google Ads Transparency Scraper
• SEEK Jobs Scraper (Seek, JobStreet, Jobsdb)
• Career Site Jobs Scraper (Workday, Greenhouse, Lever and 6 more)
• AliExpress Search Scraper
• Google News Scraper
• Sports Odds vs Prediction Markets (Kalshi, Polymarket)
• YouTube Transcripts in Bulk (whole channels, playlists)

---

Versión en español

Datos de Google Trends para muchos términos a la vez: interés en el tiempo, interés por región (países, estados, áreas metropolitanas de EE. UU., ciudades) y búsquedas relacionadas. Lee directamente los datos de Google, sin navegador, así que corre rápido y cuesta poco. Los términos sin datos en Google, y los que no se pudieron leer, se entregan gratis con el motivo. También podés poner cualquier cantidad de términos en una sola escala de 0 a 100 (el gráfico de Google llega a 5) y obtener datos diarios de hasta 5 años (Google pasa a semanales después de 270 días).

De un vistazo

• Qué hace: Google Trends para muchos términos a la vez, sin navegador: interés en el tiempo, por región, búsquedas relacionadas; cualquier cantidad de términos en una escala 0-100; datos diarios de hasta 5 años.
• Sirve para: investigación de palabras clave y mercados en volumen; comparar muchos términos en una escala; series diarias largas; interés por región.
• No sirve para: volúmenes absolutos de búsqueda (los números de Google son relativos); Trending Now; datos diarios de más de 5 años (no están en esta versión).
• Entrada: searchTerms (obligatorio; hasta 5.000 pares término × lugar); geos; timeRange (today 12-m por defecto); opcionales compareTogether + anchorTerm, dailyForLongRanges, secciones, regionResolution, category, searchType. Ejemplo mínimo en JSON: sección Minimal input de la parte en inglés.
• Salida: una fila por término y lugar (las filas sin datos van gratis, con status y el motivo) y un registro SUMMARY.
• Eventos que se cobran: keyword USD 0,003 por término con datos (USD 3,00 cada 1.000); keyword-daily USD 0,01 con dailyForLongRanges y un rango de más de 270 días (USD 10,00 cada 1.000); más apify-actor-start.
• Nunca se cobra: términos sin datos (noData), términos ilegibles (failed), el SUMMARY; no hay cargo extra de proxy ni de cómputo.
• Límites: hasta 5.000 pares por corrida; se detiene en tu costo máximo por corrida (stoppedBy).

Por qué este

• Hecho para volumen. Sin navegador: 175 términos en 5 minutos en una prueba, cada uno con su curva, sus regiones y sus búsquedas relacionadas, todos completos. Cuando Google frena la corrida, el Actor pasa solo a IPs residenciales nuevas, sin costo extra de proxy ni de cómputo.
• Pagás solo lo que tiene datos. Un término se cobra cuando vuelve su curva de interés (o, si no la pediste, cuando vuelve alguna sección). Un término sin datos en Google, o que no se pudo leer después de varios intentos, se entrega igual para que lo sepas, con status noData o failed y el motivo. No se cobra.
• Un estado por sección. Cada fila dice, para la curva, las regiones, las búsquedas relacionadas y los temas relacionados, si salió ok, empty (Google respondió sin nada), failed o notOffered. Una lista vacía nunca pasa por dato. El registro SUMMARY lista cada término que no salió completo y por qué.
• Más de 5 términos en una escala. Hasta 5 salen de un solo gráfico de Google. Con más, cada tanda de 5 lleva un término ancla que une las tandas. Prueba del 2026-10-09 con 12 comidas: los valores encadenados de empanada y falafel quedaron a menos de 0,6 % de lo que da Google con esos términos en un mismo gráfico. En el gráfico de 5 de Google con pizza, falafel sale todo en 0.
• Datos diarios de rangos largos, hasta 5 años. Cada fila trae el control contra la serie semanal de Google: 1,0 punto en 3 años y 2,6 en 5 años, en las pruebas.
• Cualquier código de ubicación que acepte Google Trends: países, regiones, metros de EE. UU. o mundial.

Precio

Pagás por término con datos: USD 3,00 cada 1.000 términos (USD 0,003 cada uno; un término en una ubicación es una fila), o USD 10,00 cada 1.000 términos con datos diarios de rangos largos (dailyForLongRanges activado y un rango de más de 270 días; USD 0,01 cada uno). Un término se cobra cuando vuelve su curva de interés (o, si apagaste la curva, cuando vuelve alguna sección). No hay cargo por corrida aparte del evento de inicio estándar de Apify, y no se cobran el registro SUMMARY, los términos sin datos en Google ni los que no se pudieron leer. Poné un cargo máximo por corrida en Apify y el Actor se detiene ahí.

Confiabilidad

Lo que se midió antes de publicar (2026-10-09), y lo que no:

• Corridas: todas las corridas de prueba terminaron. 175 de 175 términos volvieron con curva, regiones y búsquedas relacionadas; en otra corrida de 60 términos, 3 volvieron sin búsquedas relacionadas después de 8 intentos (partial, con curva y regiones). Un término que no se puede leer se entrega con status: failed y el motivo, y no se cobra.
• Bloqueos: Google rechaza una IP después de unos 25 términos seguidos. El Actor pasa solo a IPs residenciales nuevas: así volvieron 200 de 200 términos en una prueba. Si Google rechaza también esas, los términos vuelven failed (gratis), no como una corrida rota.
• Misma pregunta, mismos números: pedidos dos veces el mismo día, y con ventanas corridas, Google dio valores semanales idénticos (0 diferencias en 261 semanas). No se midió entre días distintos. Google dice que Trends trabaja con una muestra de las búsquedas y agrega "pequeñas fluctuaciones al azar" para proteger la privacidad, más visibles en términos de poco interés (Ayuda de Google Trends), así que una corrida de otro día puede diferir un poco.
• Números enteros: Google da 0-100 en enteros. En un término cuyo pico queda por debajo de 10, ese redondeo es de varios %; esas filas llevan lowPrecision.
• Todavía sin medir: la tasa de éxito en la tienda con semanas de uso real (publicado el 2026-10-09). Como referencia, el Actor de Google Trends más usado falla o se agota en más o menos 1 de cada 4 corridas (sus estadísticas públicas, leídas el 2026-10-09).

No afiliado

Sin relación con Google. "Google Trends" nombra la fuente pública de los datos.