Blog / API guides

Google People Also Ask API: Related Questions as JSON

22 Sept 2026Perceptlayer6 min read

Google has no API for People Also Ask. You can get them as JSON two ways: every GET /v1/google-search/search response includes a related_questions array, and GET /v1/google-related-questions/related-questions returns only the questions for a query q. Both cost 1 credit per successful call. Each question carries a next_page_token you pass back to fetch more.

Is there an official People Also Ask API?

No. Google doesn't offer an API for the People Also Ask box. Its general web search API is the Custom Search JSON API, and Google's own overview of it now says it is closed to new customers and that existing customers have until January 1, 2027 to move to an alternative (Google for Developers, 2026). For searching up to 50 domains, Google points to Vertex AI Search; for full web search it asks you to contact it.

So People Also Ask data comes from the results page itself. The Google Related Questions API returns the questions alone, and the Google Search API returns them with the rest of the page. Data is collected from publicly available sources and returned as found at the time of the call, and the questions change by query, language, country and day.

Where do related questions appear in a search response?

In the related_questions array of every GET /v1/google-search/search response that has a People Also Ask box. It sits next to organic_results, related_searches and people_also_search_for, so if you already run searches for rank tracking, you have the questions at no extra cost. This is the request behind the recorded response:

curl "https://api.perceptlayer.com/v1/google-search/search?query=best+restaurants+in+Berlin&hl=en&safe=off" \
  -H "X-API-Key: $API_KEY"

And the part of the response that matters here, trimmed with …:

{
  "search_information": {
    "query_displayed": "best restaurants in Berlin",
    "organic_results_count": 6
  },
  "organic_results": […],
  "related_searches": [
    { "query": "Best restaurants in Berlin for tourists", … },
    { "query": "Best affordable restaurants in Berlin", … }
  ],
  "related_questions": [
    {
      "question": "What is the coolest restaurant in Berlin?",
      "next_page_token": "fc=EqEBCmJBSmlUNHRMWjV0cUtnRWF3THBmd2poWERESG9Zd1NDdjV1eVBaTnlyQUpwWHptbXpvWEl3cEcteUc4clBmTmM2RmVqZ3gxeFBSR2Q2YnhWUXRYTV9fbU9TZ25Oa3RSM2JJQRIXZlBEYWFyWDFHc1RtN…"
    },
    {
      "question": "What is famous to eat in Berlin?",
      "next_page_token": "fc=EqEBCmJBSmlUNHRMWjV0cUtnRWF3THBmd2poWERESG9Zd1NDdjV1eVBaTnlyQUpwWHptbXpvWEl3cEcteUc4clBmTmM2RmVqZ3gxeFBSR2Q2YnhWUXRYTV9fbU9TZ25Oa3RSM2JJQRIXZlBEYWFyWDFHc1RtN…"
    },
    …
  ],
  …
}

The third question in the recording is "Where is the best place to eat in Berlin?". Note what's absent: the response holds the questions, not the answers Google shows when you expand one. Google's documentation says featured snippets can appear inside a related questions group, also known as People Also Ask (Google Search Central, 2026). If you need answer text and sources for a question, search the question itself and read its results.

How do you call the related-questions endpoint?

Send the query as q to GET /v1/google-related-questions/related-questions, with hl for language and gl for country. The response is small: the query and its related_questions, each with question and next_page_token. Use it when you want only the questions and not a full results page. This is the request behind the second recording:

curl "https://api.perceptlayer.com/v1/google-related-questions/related-questions?q=how+does+espresso+work&hl=en&gl=us" \
  -H "X-API-Key: $API_KEY"

{
  "query": "how does espresso work",
  "related_questions": [
    {
      "question": "Is espresso stronger than regular coffee?",
      "next_page_token": "fc=EqEBCmJBSmlUNHRKd1EwVllFZ2xFZ3loS2xCYmtYVEszbTQ2NEVzcDZvbExoOUltSDNaYkhpdEp5T1EyN3J0Z1pnMVA0STF3YUZ0ODV3LVQxRlBERkZTWFlpRVhSTmJLd2x1YWhvZxIXVHNuYmFxN3JIZGZoN…"
    },
    { "question": "Do you sip or chug an espresso?", "next_page_token": "fc=EqEBCmJBSmlUNHRKd1EwVll…" },
    { "question": "Is an espresso 1 or 2 shots?", "next_page_token": "fc=EqEBCmJBSmlUNHRKd1EwVll…" }
  ]
}

All four parameters are optional in the endpoint definition, but you need either q or a next_page_token for the call to mean anything. Keys go in the X-API-Key header; get a free API key and see the authentication docs.

What does next_page_token actually do?

It continues the question list for the query you asked, not for one question. In both recorded responses, the Berlin search and the espresso call, every question carries an identical next_page_token. A token that's the same for all three questions can't point at one of them, so treat it as a page cursor: pass it back to the related-questions endpoint to ask for the next batch of questions for the same query.

curl -G "https://api.perceptlayer.com/v1/google-related-questions/related-questions" \
  --data-urlencode "next_page_token=$TOKEN" \
  -d hl=en -d gl=us \
  -H "X-API-Key: $API_KEY"

Tokens run up to 8,192 characters and contain characters such as =, so URL-encode them, as --data-urlencode does here. We haven't recorded a token follow-up, so check what comes back in the playground on the related-questions endpoint page before you build a loop that depends on its shape.

To go deeper on a particular question, which is what most keyword research needs, send that question as a new q. Its own related questions are the next level of the tree.

How do you build a question tree for a keyword?

Start from a seed keyword, collect its questions, then call the endpoint again with each question as q, down to a fixed depth. Keep a set of questions you've already seen, because related questions repeat across branches, and a credit cap so the tree can't grow past your budget. This script does a breadth-first walk and stops at either limit:

import os
import requests

URL = "https://api.perceptlayer.com/v1/google-related-questions/related-questions"
HEADERS = {"X-API-Key": os.environ["API_KEY"]}

def question_tree(seed, max_depth=2, max_calls=20):
    seen, edges, calls = {seed.lower()}, [], 0
    level = [seed]
    for depth in range(max_depth):
        next_level = []
        for q in level:
            if calls >= max_calls:
                return edges, calls
            r = requests.get(URL, params={"q": q, "hl": "en", "gl": "us"},
                             headers=HEADERS, timeout=60)
            if r.status_code != 200:
                print("skip", q, r.json()["error"]["code"])
                continue
            calls += int(r.headers.get("X-Credits-Used", "0"))
            for item in r.json().get("related_questions", []):
                text = item["question"]
                if text.lower() not in seen:
                    seen.add(text.lower())
                    edges.append((q, text, depth + 1))
                    next_level.append(text)
        level = next_level
    return edges, calls

edges, credits = question_tree("how does espresso work")
for parent, child, depth in edges:
    print(depth, child, "<-", parent)
print("credits used:", credits)

We haven't run this script live for this article. The counter adds X-Credits-Used from each response, so a failed call adds 0 and the total matches what your balance shows. The credits docs explain that header and X-Credits-Remaining.

What does it cost per keyword?

Each successful call to either endpoint is 1 credit, so the cost is the number of calls. If you already search the keyword for rankings, its first level of questions is free. Otherwise it's 1 call. Each question you expand is one more call. With three questions per response, as in both recordings:

Each level multiplies the calls by the number of questions, so cap the depth or the calls. Two levels for 200 seeds, 800 credits, fits in the 1,000 free credits a month; three levels at 2,600 credits fits in a Starter pack of 10,000. Deduplication only lowers these numbers. The pricing page has what packs cost.

For tracking how a keyword's questions change over time alongside rankings, the guide to SERP data for SEO monitoring stores them per day. To try the endpoint, run your own keyword in the playground on the Google Related Questions API page.

people also askgoogleseokeyword researchrelated questions