Skip to content
Search API · Use case

Find the queries people type with a Google Autocomplete API

A Google Autocomplete API returns the queries people start typing, the most direct record of how a market phrases a problem. SEO teams plan content with it, product teams name features with it, and agents expand a vague prompt into precise searches. The Search API returns suggestions, related searches and People Also Ask questions as JSON.

The problem

Keyword lists come from guesses, or from a tool that guesses for you

Content that ranks answers the question people actually type. A brainstorm produces the phrasing a team uses internally, which rarely matches the long-tail queries buyers enter, and those long-tail variants are where a new page has a chance.

Typing seeds into a search box and copying the dropdown works for five queries, not five hundred. Keyword suites solve scale with their own databases and a seat price, and calling a suggest endpoint yourself means handling blocking and parsing for a result that is only a list of strings.

type: autocomplete returns suggestions as value strings, and it is lightweight and fast. The default web search adds relatedSearches and peopleAlsoAsk to the results page, turning one seed into follow-up queries and real questions. location geo-targets both by country, so the ideas match the market you write for.

How it works

How to get keyword suggestions from the Google Autocomplete API

Expand a seed with autocomplete, add a letter to reach the long tail, then pull the questions from the results page. The autocomplete guide shows the result shape.

1 · Expand a seed
import createClient from 'microlink.io'

const microlink = createClient({
  apiKey: process.env.MICROLINK_API_KEY
})

const page = await microlink.search('headless browser', {
  type: 'autocomplete',
  location: 'us'
})

const ideas = page.results.map(({ value }) => value)

Each suggestion is a value string. Autocomplete results have no url, so there is nothing to expand: they are inputs for the next query.

2 · Reach the long tail with a letter
import createClient from 'microlink.io'

const microlink = createClient({
  apiKey: process.env.MICROLINK_API_KEY
})

const seed = 'headless browser'
const letters = 'abcdefghijklmnopqrstuvwxyz'.split('')

const pages = await Promise.all(
  letters.map(letter =>
    microlink.search(seed + ' ' + letter, {
      type: 'autocomplete',
      location: 'us'
    })
  )
)

const longTail = Array.from(
  new Set(pages.flatMap(({ results }) => results.map(({ value }) => value)))
)

Twenty-six requests collect the suggestions for every next letter, deduplicated with a Set. Save it for the seeds that matter, not every idea on the list.

3 · Pull questions and related searches
import createClient from 'microlink.io'

const microlink = createClient({
  apiKey: process.env.MICROLINK_API_KEY
})

const serp = await microlink.search('headless browser api', {
  location: 'us'
})

const questions = (serp.peopleAlsoAsk || []).map(({ question }) => question)
const related = (serp.relatedSearches || []).map(({ query }) => query)

The default search type returns peopleAlsoAsk and relatedSearches next to the results when Google shows them. Questions become headings and FAQ entries; related searches become the next seeds.

Parameters used
  • type 'autocomplete' returns suggestions as value strings. Results cannot be expanded.
  • location Two-letter country code. There is no language option, only country.
  • relatedSearches Follow-up queries from the web results page, as objects with a query field.
  • peopleAlsoAsk Related questions with question, snippet, title and link, when Google shows them.

None of these surfaces returns search volume, cost per click or keyword difficulty. Use them to discover phrasing and questions, then size the shortlist with the metrics source you already trust. The query expansion pattern shows autocomplete feeding heavier searches.

Why it works

Why autocomplete beats a brainstorm for keyword research

Suggestions are what people type, in the words they type it. That makes them the right raw material for a keyword list.

01 · Demand in its own words
Suggestions are real queries, not synonyms.
Autocomplete completes a prefix with queries people search for, which is why its phrasing fits titles and headings better than a thesaurus. The guide lists demand modeling and prompt seeding among its uses.

Check where you already rank for the phrases you find with rank tracking by country.

02 · Questions included
People Also Ask arrives structured.
peopleAlsoAsk comes back as question, snippet, title and link, so an FAQ outline is a map over an array. The link shows the source behind each answer, which is the page your content competes with.

Want the whole results page as text for a content brief? Get the SERP as Markdown from the same query.

03 · Per market
location sets the country.
Pass a two-letter location code to research the market you write for rather than the one your office sits in. Every seed and every letter variant is one request, so a market is a known cost.

When not to: autocomplete does not tell you how many people search a phrase or how hard it is to rank for it. If you prioritize by volume or difficulty, pair these ideas with a keyword metrics source. Grounding an assistant in live results instead? See live search for LLM answers.

FAQ

How do I use the Google Autocomplete API for keyword research?

Call microlink.search with a seed and type: 'autocomplete'. Each result is a value string with a suggested query. Append letters to the seed to reach long-tail variants, and run the default search type to add related searches and People Also Ask questions.

Does the keyword suggestions API return search volume?

No. Autocomplete returns suggested queries only, and web results add related searches and questions. There is no volume, cost per click or difficulty field, so size your shortlist with a metrics source.

Can I get People Also Ask questions through the Search API?

Yes. The default search type returns peopleAlsoAsk with question, snippet, title and link when Google shows the box for that query, alongside relatedSearches and the organic results.

How many requests does autocomplete keyword research use?

One per query. A seed plus 26 letter variants is 27 requests, and each web search for questions is one more. Search has no free tier: it is paid from the first request, with Pro plans from €39/month for 46,000 requests.

Is this Google Suggest API an official Google product?

No. Microlink Search is an independent product that queries public Google surfaces, autocomplete included. It is not affiliated with or endorsed by Google, and Google is a trademark of Google LLC.
Related use cases

Solve the next problem with the same API

Google rank tracking by country

Compute where a domain ranks for each keyword in each country from ordered results, and keep the history in your own database.

Google results page as Markdown or HTML

Start from a query, not a URL: get the structured results plus the Google results page itself as Markdown or HTML.

Live search to ground LLM answers

Retrieve fresh results for a question, read the best sources as Markdown and pass them to the model with citations.

Local business leads from Google Maps

Turn a category and a city into a lead list with phone, website, hours, rating and Place ID, then pull emails from each website.

LLM context from any URL

Compose Markdown, links, emails, metadata and tech stack from one URL into a context object for your agent.

Custom fields alongside metadata

Get prices, ratings, headings or any CSS selector, typed and returned next to the normalized metadata.

Ready to find what people search?

Autocomplete suggestions, related searches and questions as JSON, per country. Get a Pro key and expand your first seed today.