Google People Also Ask Scraper - Question Tree

This Google People Also Ask scraper collects the questions Google shows in its "People also ask" box for your keywords, then expands each question into a tree. Every row carries the original seed, the question that produced it, its depth and its position. This makes the dataset useful for content briefs, topic research and FAQ planning without losing the relationships between questions.

Run it from the Apify Console, the API, a schedule or an integration. Give it a list of keywords, a language and a country, and you get back a flat dataset of distinct questions that you can export as JSON, CSV or Excel, or rebuild into a tree with the seed, parent_question and depth fields.

Input

Provide keywords, one keyword per list entry. The existing default input is ["web scraping"]. The actor also accepts comma-separated or newline-separated text where the input interface permits text. Whitespace and empty entries are removed. There is no required field in the input form, but a run with no nonempty keywords returns an empty dataset and a bad_input entry in the errors record.

depth defaults to 1. Zero reads only the seed's own People Also Ask box. One also searches each question from that box and collects its related questions. Two and three continue the same expansion, up to the maximum depth of three. Setting zero deliberately keeps the run flat. More depth creates more searches and can take longer.

max_questions defaults to 200 and permits 1 through 5000. It is a total limit for the whole run. Seeds are processed in the order supplied, so an earlier seed may use the entire limit before later seeds are searched. Reduce depth or increase the limit when you want to explore more of a multi-keyword input.

language defaults to en and selects Google's interface language. region defaults to us and selects the Google market. Both are normalized to lowercase and remain visible on every result. attempts defaults to 20 and permits up to 60 attempts per query. It controls how often a query can be retried when Google does not return a readable search page. Large trees continue automatically; no manual page handling is needed when running this actor.

This is the actual input used for the saved default measurement:

{
  "keywords": [
    "web scraping"
  ],
  "depth": 1,
  "max_questions": 200,
  "language": "en",
  "region": "us",
  "attempts": 20
}

Google People Also Ask output

The default dataset contains one row per distinct question. question is the question shown by Google. seed is the original keyword and stays the same for its descendants. parent_question is the search that produced this question: the seed at depth zero, and a previously found question at deeper levels. It is always a string.

depth describes the expansion level and position runs from one through the number of returned questions. language and market preserve the normalized input options. The added fields source, engine, query, fetched_at and country provide explicit source identification, retrieval time and market information. query is the same search as parent_question; the original fields remain available for existing integrations.

Here is one actual result from the saved default run:

{
  "question": "Can Chatgpt do web scraping?",
  "seed": "web scraping",
  "parent_question": "web scraping",
  "depth": 0,
  "language": "en",
  "market": "us",
  "position": 1,
  "source": "google-people-also-ask",
  "engine": "google",
  "query": "web scraping",
  "fetched_at": "2026-10-05T01:23:04.343Z",
  "country": "US"
}

A deeper run shows how the tree grows. This is part of the actual output for the Dutch keywords elektrische fiets and zonnepanelen with depth 2, max_questions 40, language nl and region nl. The second row was found by searching the first one:

[
  {
    "question": "Wat zijn de 10 beste elektrische fietsen?",
    "seed": "elektrische fiets",
    "parent_question": "elektrische fiets",
    "depth": 0,
    "language": "nl",
    "market": "nl",
    "position": 1
  },
  {
    "question": "Wat is de meest betrouwbare elektrische fiets?",
    "seed": "elektrische fiets",
    "parent_question": "Wat zijn de 10 beste elektrische fietsen?",
    "depth": 1,
    "language": "nl",
    "market": "nl",
    "position": 6
  }
]

The key-value store's summary record reports the number of seeds, searches sent, questions returned, questions per depth and queries that produced no questions. It also reports retries and errors. If a query cannot be read, its blocked error is recorded while other readable queries continue. Empty search results have no question rows. No error record is written for an ordinary search that simply has no questions. The errors record is present when there are errors.

Google loads answers when a question is opened in the browser. The server-side search page used here provides questions, so this actor returns questions without answers. The summary explicitly reports answersAvailable: false. An empty result is possible for narrow or unusual keywords; it does not imply that questions or answers were invented.

Use cases for People Also Ask data

• Content briefs. Turn the questions behind a keyword into H2s and sections, so an article answers what searchers actually ask next.
• FAQ pages and schema. Pick the questions closest to your product and answer them on the page that should rank for them.
• Topic clusters. The parent and depth fields show which questions belong together, which helps decide what deserves its own page and what belongs in an existing one.
• Market and language comparison. Run the same keyword for different countries or languages and compare which questions people ask in each market.
• Monitoring. Schedule the same input weekly and track how the People Also Ask box for your important keywords changes over time.

How this compares to other People Also Ask tools

Copying questions by hand from the search page works for one keyword, but each click opens more questions and the relationships are lost as soon as you paste them into a document. Browser extensions help for a single search but do not run on a schedule or over a list of keywords. Subscription keyword suites and dedicated People Also Ask tools show a visual tree, usually with a monthly plan and limits on searches per month.

This actor is a pay-per-use People Also Ask tool: no subscription, a small run fee and a price per question delivered. It returns structured rows that fit a spreadsheet, a database or an automation, and it covers any Google language and country you specify. It does not return answers or source links; if you need answer text, combine the questions with your own research or another tool.

Pricing

The existing price is $0.002 per run start and $0.003 per question delivered. The run start is charged once, including runs with empty results or invalid keyword input. No question event is charged when no questions are delivered. The saved standard run returned 13 questions, corresponding to one run start and 13 question events. The saved sourdough run returned 17 questions, and the Dutch depth-two input returned its limit of 40 questions. Actual question counts can change because Google's suggestions change.

Free-plan runs deliver at most ten questions. The same question price applies only to those delivered rows, and the run start remains unchanged. max_questions limits the number of question events, which helps bound the requested output volume.

FAQ

Does this actor return answers?

It returns the questions Google publishes on the search page. Answers are not included. Use parent_question and depth to reconstruct the relationships between questions.

Why do repeated runs return different questions?

Google changes the People Also Ask box over time and across markets. The same input can therefore produce different wording or question counts. Language and market fields help distinguish results collected for different search settings.

Why did a later keyword receive no results?

The question limit is global and seeds run in input order. If the first seed reaches max_questions, the run stops. Use a smaller expansion depth or separate runs when each seed needs its own output allowance.

What happens when Google does not answer a query?

The query is retried up to the requested attempts. If it remains unreadable, a blocked entry is recorded, other queries continue, and no question rows are charged for that failed query. A complete lack of readable queries produces a successful empty actor run.

Can I use this as a People Also Ask API?

Yes. Start the actor through the Apify API with the same input as in the Console and read the dataset when the run finishes. Every Apify integration, such as webhooks, Make, Zapier or a scheduled run, works the same way.

Which languages and countries are supported?

Any two-letter Google interface language in language and any two-letter country code in region. The questions come back in the requested language for that market, as shown in the Dutch example above.