Google Trends Scraper — Fast HTTP, Failed Queries Free

Scrape Google Trends data over pure HTTP — no browser. Interest over time, interest by region, related queries and related topics for any keyword, location and time range. Compare up to 5 terms, batch keywords from a Google Sheet, export flat CSV-ready rows. Failed and partial queries are free.

Pay only for complete successful queries. Partial and failed queries have no Actor event fee; Apify platform usage is charged separately, including unsuccessful attempts.

Google Trends Scraper at a glance

This is an independent Google Trends scraper by joshua yan on Apify. Its stable Actor identifier is joshuayan/google-trends-scraper. It collects Google Trends web data over HTTP and can be run from Apify Console or the Apify API.

Question	Answer
What data does it return?	Interest over time, geographic interest, related queries, and related topics when Google supplies them.
What inputs does it accept?	Keyword lists, Google Trends explore URLs, or the first column of a public Google Sheet.
Can it compare keywords?	Yes, up to five search terms in one comparison query.
What can I export?	JSON query results or flat data rows suitable for CSV and spreadsheets.
What do the numbers mean?	Relative Google Trends interest, not absolute monthly search volume.
What is charged?	One event per complete successful query, including complete no-data responses. Partial and failed queries have no event fee; platform usage is separate. See pricing.
What is the compatibility status?	Full output compatibility with the official Apify Actor remains unverified. Validate downstream integrations before migrating production workloads.

Try the Actor · Input fields · API examples · Report an issue

SEO, content research, and GEO planning use cases

• SEO keyword research: compare relative interest in candidate search terms, inspect seasonality, and review top or rising related queries. Keep geography and date ranges consistent when comparing results.
• Content planning: use related queries and topics to build a research shortlist, then check the underlying results before choosing articles, FAQs, or campaign topics.
• Regional market research: compare geographic interest to help prioritize markets. Geographic coverage and resolution depend on the data Google returns.
• GEO topic research: use search-interest signals as one input to generative engine optimization content planning. This Actor does not query AI answer engines or measure brand mentions, citations, share of voice, or AI search rankings.
• Automated research workflows: call the Actor through the Apify API and pass JSON or flat rows to analysis tools. Review partial-result indicators and the run summary before using the output downstream.

What this Google Trends Scraper does

Collect interest over time, geographical interest, related queries, and related topics. Use keyword lists, Google Trends URLs, or the first column of a public Google Sheet. Compare up to five terms in one query. Export either one compatible item per query or flat rows for CSV and spreadsheets.

Results are relative Google Trends interest values, not absolute search volumes. Missing optional sections produce a partial result. A query whose curve cannot be retrieved is recorded in FAILED_QUERIES without a result item.

HTTP collection and measured performance

This Actor uses HTTP requests and a reusable session pool, so it does not launch a browser for each query. It also exposes missing sections and supports flat output. Partial results are delivered without an event fee. Exact compatibility across all modes still requires the complete official baseline before production migration.

On October 22, 2026, private run duxbgu9yKcWSx0P1x completed 100 distinct queries: 34 worldwide keywords, 33 US URLs, and 33 three-term comparisons. All 100 were complete, with no partial or failed queries in this sample. At 256 MB using US residential exits, P95 was 6.588 seconds including output, state persistence, and charging; fetching alone was 5.439 seconds. The run made 569 HTTP requests. This is one observed sample, not an availability guarantee.

The official ten-query comparison delivered only one result, so it cannot establish a valid cost or reliability comparison.

Migrating

The input field names, titles, and selectable values were taken from the official Actor's public build 0.0.141. Keyword lists and Google Trends URLs can be reused. The browser options skipDebugScreen, maxConcurrency, maxRequestRetries, and pageLoadTimeoutSecs are accepted and ignored. Query concurrency, retry policy, and timeout are managed by this Actor.

A live US weather URL has passed a same-query official output shape comparison. Comparison-mode related results currently use a term field, which still needs an official comparison baseline: repeated official attempts returned no usable comparison output. Verify downstream consumers before production migration. missingSections is an additional field; relatedTopics_rising also exists in the observed official output.

Input

Provide at least one source: searchTerms, startUrls, or spreadsheetId.

Field	Meaning
searchTerms	Array of keyword strings. Surrounding whitespace and empty entries are removed when queries are expanded.
isMultiple	Default false. When true, comma-separated terms in each string form one comparison query, with at most five nonempty terms.
startUrls	Google Trends explore URLs as { "url": "..." } objects. URLs use their own filters. Missing date, geo, and category default to the past 12 months, worldwide, and all categories; they do not inherit keyword filters.
spreadsheetId	ID of a public Google Sheet. Reads the first column, skips the header, and ignores blank cells.
timeRange	Official presets: empty/default (past 12 months), now 1-H, now 4-H, now 1-d, now 7-d, today 1-m, today 3-m, today 5-y, or all.
customTimeRange	YYYY-MM-DD YYYY-MM-DD, with real dates and start no later than end. Overrides the preset for keyword and Sheet queries.
geo	Data geography; empty means worldwide. A URL may specify a country or subregion.
category	Category selected in the input form, sent as a string such as "5" (Computers & Electronics). Empty means all categories. A Trends URL can specify other category IDs using its cat parameter.
outputMode	compatible (default) or flat.
maxItems	Maximum number of queries that produce output. 0 means unlimited. In flat mode this limits queries, not rows.
viewedFrom	Country code for the Apify residential proxy IP; case-insensitive, defaults to US. Independent of data geography.

Single keyword:

{ "searchTerms": ["webscraping"], "maxItems": 10 }

Comparison:

{
  "searchTerms": ["chatgpt,gemini,claude"],
  "isMultiple": true,
  "timeRange": "today 3-m",
  "outputMode": "flat"
}

From a URL:

{
  "startUrls": [
    { "url": "https://trends.google.com/trends/explore?q=coffee&geo=US&date=today%203-m" }
  ]
}

Output

Compatible

One item per complete or partial query. The following illustrates the structure and is not a measured live result:

{
  "inputUrlOrTerm": "webscraping",
  "searchTerm": "webscraping",
  "interestOverTime_timelineData": [],
  "interestOverTime_averages": [],
  "interestBy": [],
  "interestBySubregion": [],
  "interestByCity": [],
  "interestByMetro": [],
  "relatedQueries_top": [],
  "relatedQueries_rising": [],
  "relatedTopics_top": [],
  "relatedTopics_rising": [],
  "missingSections": []
}

All four geography arrays are present. COUNTRY uses interestBy; REGION populates interestBySubregion and an additional CITY request populates interestByCity. CITY uses interestByCity; DMA uses interestByMetro, verified with a California URL. Unused arrays are empty.

Timeline points include time, formattedTime, formattedAxisTime, value, hasData, and formattedValue, preserving optional isPartial when Google returns it. Related entries contain a query or topic, values, and term for the source comparison term. missingSections contains missing optional section names: region, relatedQueries, or relatedTopics. The curve is required for output.

Flat

One row per data point, with 27 columns (schemaVersion: 2). The original 11 columns remain first in the flat dataset view and CSV export: queryId, searchTerm, section, listType, time, geoCode, geoName, label, value, formattedValue, and isBreakout.

The following 16 columns are appended:

Columns	Meaning
sourceTerm	Source keyword for the row; null when attribution is unavailable.
topicId, topicType	Google topic identity and type; null for non-topic rows or unavailable data.
geoResolution	COUNTRY, REGION, CITY, or DMA for region rows; otherwise null.
hasData, isPartial	Upstream data availability and timeline bucket quality. isPartial describes a time bucket, not the whole query.
schemaVersion, source	2 and google-trends-web.
inputUrlOrTerm, geo, timeRange, category, property	Effective query identity and filters.
retrievedAt	UTC retrieval timestamp, persisted with the rows and preserved on recovery.
queryStatus, missingSections	Existing query processing status and detected missing sections. These do not guarantee that Google supplies nonempty related topics.

section is timeline, region, relatedQuery, or relatedTopic; listType is top, rising, or null. The default compatible output and per-query event fee are unchanged. More columns increase export size and can add storage and transfer usage; flat rows do not incur an extra Actor event fee. Select the Flat rows dataset view, or use the dataset API with format=csv&view=flat, to export the ordered columns.

Example timeline row:

{
  "queryId": "q0",
  "searchTerm": "weather",
  "schemaVersion": 2,
  "source": "google-trends-web",
  "inputUrlOrTerm": "weather",
  "geo": "US",
  "timeRange": "2026-07-01 2026-09-30",
  "category": 0,
  "property": "",
  "retrievedAt": "2026-10-23T08:49:18.930Z",
  "queryStatus": "succeeded",
  "missingSections": [],
  "section": "timeline",
  "listType": null,
  "time": "1782864000",
  "geoCode": null,
  "geoName": null,
  "label": "weather",
  "value": 71,
  "formattedValue": "71",
  "isBreakout": false,
  "sourceTerm": "weather",
  "topicId": null,
  "topicType": null,
  "geoResolution": null,
  "hasData": true,
  "isPartial": false
}

Run summary and recovery

The OUTPUT key contains totalQueries, succeeded, partial, failed, charged, and stoppedBy (null, maxItems, chargeLimit, or chargeUncertain). FAILED_QUERIES contains queryId, input, reason, and attempts.

Each query's prepared result and dataset position are persisted before output. Resuming the same run checks the existing rows and appends only a missing suffix, including interrupted flat batches. Already fetched results are reused. Keep the same query inputs and output mode; changed inputs are rejected. Do not append to the run's dataset from another writer while it is running or being resumed. Conflicting rows stop recovery instead of being overwritten.

A per-query charging ledger records the intent and receipt. A completed receipt is reused on resume; a delivered query whose charge was not attempted can finish charging without another fetch. If a charge response is lost, the Actor checks the run's charged-event counter. When confirmation remains ambiguous, it stops with stoppedBy: "chargeUncertain", retains pendingCharge in RUN_STATE, and does not send another charge. Resume the same run to retry confirmation. A stop before sending the charge can also be ambiguous; automated recovery cannot guarantee a charge in that case. CHARGE_<queryId> records the reconciliation state. Legacy runs without these journals retain their earlier recovery limits.

Pricing

Base price: $2 per 1,000 complete queries ($0.002 per query), plus Apify platform usage costs. There is no separate Actor start fee or flat-row event fee.

Introductory plan discounts

Introductory discounts took effect on October 23, 2026. We plan to review these rates after the initial 60-day period. Check the live Pricing tab for the price that applies to your plan before starting a run.

Discount tier	Discount from base price	Per complete query	Per 1,000 complete queries
FREE	0%	$0.002	$2.00
BRONZE	60%	$0.0008	$0.80
SILVER	80%	$0.0004	$0.40
GOLD	87.5%	$0.00025	$0.25
PLATINUM (enterprise)	87.5%	$0.00025	$0.25
DIAMOND (enterprise)	87.5%	$0.00025	$0.25

Enterprise tiers currently use the same event price as GOLD. These are event fees only. Any later price increase will follow Apify's notification period; the offer does not automatically revert on day 60.

The 100-query benchmark consumed $0.05708758 of platform usage. At the undiscounted base event rate, the same usage would total approximately $0.25709 for 100 complete queries, or $2.571 per thousand. This is a projection: actual proxy, storage, compute, and retry costs vary by workload and Apify plan. The earlier benchmark used a lower private test event price; it is not the public event price. The incomplete official run is excluded from valid unit-cost comparison.

One trends-query event is charged for each complete successful query, including a complete response with no data. Partial results, failed queries, and input validation failures do not charge this event. charged in the summary counts queries with an accepted event charge.

Apify platform costs, including residential proxy traffic and compute, are separate from the event fee and may be incurred by unsuccessful attempts. Once the event budget is reached, no new queries are claimed; queries already running can finish and produce uncharged output. Local SDK test pricing is a simulation, not the production price.

FAQ

Is this the official Google Trends API? No. This is an independent community Actor that collects data from Google Trends web endpoints. Apify provides the API used to run the Actor. It is not an official Google product or a complete drop-in replacement for the official Apify Google Trends Actor.

Can I get Google Trends data as CSV? Yes. Choose outputMode: "flat" to produce one row per data point, then export the dataset as CSV. A query can produce many rows; the event fee is per complete query, not per exported row.

Does this measure monthly keyword search volume? No. Results are relative Google Trends interest values. They should not be presented as exact search counts, forecasts, or advertising keyword volumes.

Can this track visibility in ChatGPT, Gemini, or other AI answers? No. It can supply topic-demand research for a GEO workflow, but it does not retrieve AI answers or measure their mentions and citations. Its geo input means data geography, not generative engine optimization.

Are failed Google Trends queries completely free? They have no Actor event fee, and partial results also have no event fee. Apify compute, proxy, storage, and other platform usage can still be charged for these attempts.

How are 429 responses handled? The session is retired and requests across the session pool share a cooldown. Cooldown increases from 5 to at most 30 seconds, also considering Google's Retry-After header within that bound. HTML block pages and Google consent/sorry redirects receive the same treatment. Requests allow up to three retries; JSON parse failures allow one retry. An optional widget that still fails with a transient or rate-limit error gets one final recovery attempt, without fetching the curve or successful widgets again. Persistent failures produce a failed or partial query. Waiting and retries can increase latency; this does not guarantee freedom from rate limits.

Can I compare more than five terms? No. Split them into groups of at most five. Each group counts as one query and its timeline contains one value per term.

Why are geography arrays empty? All four fields are retained for compatibility; only the geographic resolution Google supplies is populated. This Actor requests CITY as well when Google supplies a REGION widget. A US-CA URL may return DMA/metro data.

Why is my Sheet rejected? It must be publicly readable without signing in and contain a header plus at least one nonblank value in the first column. Quoted CSV cells, commas, and embedded newlines are supported.

Can I use custom proxies? By default, the Actor uses Apify residential proxies. TRENDS_PROXY=none enables direct local debugging. A custom HTTP(S) proxy URL is supported through TRENDS_PROXY; viewedFrom only controls Apify residential proxies.

Is this a drop-in replacement for the official Apify Actor? No. Private performance, real response parsing, and one US same-query official shape comparison are verified. Complete official-output compatibility and a valid ten-query cost comparison remain pending. In particular, comparison-mode related-item formatting still needs an official baseline.