Google Related Searches Scraper

Collects live query links from Google's native related-search modules. HTTP extraction runs first; an optional browser fallback reads the same public result page. Queries are processed with concurrency two and output has one global limit.

Input

Supply queries (an array) or query, language (default en), country (default us) and maxResults (default 100, maximum 10000). The sample requests ten terms across javascript tutorial, python tutorial and typescript tutorial. A single Google module often contains fewer than ten terms. Results are never padded.

Use proxyConfiguration for HTTP requests. The sample uses GOOGLE_SERP. browserFallback defaults true; its optional browserProxyConfiguration defaults to residential. Custom proxies are supported. Search availability and proxy use can incur platform charges.

Output and provenance

Each row contains id (derived seed plus term), seedQuery, relatedSearch, source rank, actual search url, visible linkText, sourceModule and optional heading. Metadata records the requested source URL, provider, extraction method and retrieval time. Unknown fields are omitted. Rank is within the observed module, not an organic ranking.

Only the native #bres container or a module with an exact Related searches / People also search for heading is accepted. Heading fallback currently recognizes English. Autocomplete, People also ask and AI Overview notices are excluded. The actual HTML is retained in run key-value storage for source review. No CAPTCHA solving or authenticated-session reuse is performed.

Limits and verification

The ten-record local sample used English US tutorial queries and real #bres links. Other languages and query combinations depend on Google's layout and availability. Duplicate terms within a seed are removed; the same term under a different seed retains separate provenance. A blocked or unavailable query is reported in RUN_SUMMARY. A run with no genuine terms fails explicitly; partial results remain partial.

Run

Use apify run --input-file input.json locally. Export the default Dataset after a successful run; inspect RUN_SUMMARY for query errors and requested versus actual counts.