Google, Bing & YouTube Keyword Suggest Scraper

Turn seed keywords into hundreds of long-tail ideas

Generate structured keyword suggestions from Google, Bing, and YouTube autocomplete for SEO, PPC, content planning, marketplace research, and demand discovery.

Start with up to 100 seed keywords, expand them through A–Z, digits, and custom modifiers, target a language and country, and export deduplicated long-tail suggestions into a clean Apify dataset.

No API key. No login. No browser automation. No LLM.

Key features

• Google, Bing, and YouTube autocomplete
• Up to 100 seed keywords per run
• A–Z long-tail expansion
• 0–9 digit expansion
• Custom prefix and suffix modifiers
• Language and country targeting
• Direct suggest mode
• Automatic deduplication per source
• Structured keyword output
• Free RUN_SUMMARY record on every run
• Built-in source health and contract checks
• Schedule-ready for recurring keyword research
• Pay only for keyword records actually written

Unofficial community Actor. Not affiliated with, sponsored by, or endorsed by Google, YouTube, or Microsoft. Data comes from public autocomplete endpoints and remains subject to the respective providers' terms.

---

What this Actor is for

Autocomplete suggestions can reveal useful search phrasing around a topic without requiring you to manually type hundreds of variations into search boxes.

Use this Actor to expand a small seed list into structured long-tail keyword ideas you can analyze, filter, export, or send into another workflow.

Typical use cases include:

• SEO keyword discovery
• long-tail content research
• PPC keyword ideation
• buyer-intent research
• localized keyword research
• marketplace and product research
• topic expansion
• content calendar planning
• recurring autocomplete monitoring
• competitive or market research

---

Who it's for

SEO and content teams

Turn a topic into a larger long-tail dataset.

For example:

crm software

can lead to suggestions around more specific needs, comparisons, audiences, and use cases.

Use the results to support:

• content briefs
• landing-page research
• FAQ discovery
• topical maps
• editorial planning
• long-tail SEO analysis

PPC and performance marketing teams

Discover additional search phrasing around products, services, competitors, problems, and purchase intent.

Use country and language targeting to research different markets separately.

E-commerce and marketplace sellers

Expand product or category seeds into more specific phrases that can support:

• listing research
• campaign ideation
• category analysis
• product positioning
• buyer-language research

Market researchers

Collect autocomplete suggestions as demand and phrasing signals around a topic.

This Actor does not claim that autocomplete suggestions are search-volume measurements. Use a separate volume source when exact search-volume estimates are required.

---

Supported sources

The Actor supports:

• google
• bing
• youtube

You can run one source or multiple sources in the same execution.

Example:

{
  "sources": ["google", "bing"]
}

Using multiple sources provides broader keyword coverage and also gives a recurring workflow some resilience if one public autocomplete route changes.

---

Two operating modes

Expand mode

expand is the default mode and is designed for long-tail harvesting.

Each seed is queried directly and expanded with:

a ... z
0 ... 9

So a seed such as:

crm software

generates query variants such as:

crm software
crm software a
crm software b
crm software c
...
crm software z
crm software 0
crm software 1
...
crm software 9

This creates approximately 37 base queries per seed before custom modifiers are added.

Autocomplete responses vary by topic, source, market, and time, so the number of unique keywords is not guaranteed. A single seed can nevertheless generate hundreds of distinct long-tail suggestions.

Suggest mode

suggest performs one direct autocomplete query per seed without A–Z or digit expansion.

Example:

{
  "seeds": ["crm software", "email marketing"],
  "mode": "suggest"
}

Use this mode when you want:

• direct autocomplete suggestions
• fewer requests
• faster focused checks
• recurring monitoring of a fixed seed list

---

Custom modifiers

You can add up to 20 custom modifiers.

Example:

{
  "modifiers": ["best", "how", "vs"]
}

Modifiers are applied as both prefixes and suffixes.

For a seed such as:

crm software

the Actor can generate variants such as:

best crm software
how crm software
vs crm software
crm software best
crm software how
crm software vs

This is useful for expanding research around comparison, commercial, informational, or custom niche terms.

---

Input

Long-tail harvest example

Google + Bing:

{
  "seeds": ["crm software", "email marketing"],
  "mode": "expand",
  "sources": ["google", "bing"]
}

Localized research example

{
  "seeds": ["crm", "automação de vendas"],
  "mode": "expand",
  "sources": ["google", "bing"],
  "language": "pt",
  "country": "br",
  "modifiers": ["melhor", "preço", "para empresas"]
}

Input fields

Field	Type	Default	Description
seeds	array	required	1–100 seed keywords.
mode	string	expand	expand for A–Z/digit long-tail harvesting or suggest for direct autocomplete only.
sources	array	["google"]	Supported sources: google, bing, youtube.
language	string	en	Language code, such as en, pt, or de.
country	string	us	Country code, such as us, br, or de.
modifiers	array	[]	Up to 20 custom terms applied as both prefix and suffix.
maxResults	integer	5000	Maximum number of charged keyword records written. The free summary does not count toward this cap.
requestDelayMs	integer	250	Delay before each request. Minimum accepted value is 100 ms.

---

Why Google + Bing is useful for recurring runs

Public autocomplete routes are useful but undocumented.

A provider can change its response shape without notice.

When multiple sources are selected, the Actor handles them independently. If one source produces an incompatible response and is flagged with:

API_CONTRACT_CHANGED

the Actor can continue processing the other selected source instead of automatically turning the entire run into an empty result.

This is especially useful for scheduled research workflows.

The product default remains Google when sources is omitted.

---

Output

The dataset contains one record per keyword.

Keyword records use:

recordType: "KEYWORD"

Example keyword record

{
  "recordType": "KEYWORD",
  "keyword": "crm software for nonprofits",
  "seed": "crm software",
  "variantQuery": "crm software f",
  "position": 3,
  "source": "google",
  "language": "en",
  "country": "us",
  "detectedAt": "2026-07-30T13:00:00.000Z"
}

Keyword fields

Field	Description
recordType	Record type. Keyword rows use KEYWORD.
keyword	Autocomplete suggestion returned by the selected source.
seed	Original seed keyword that produced the result.
variantQuery	Expanded or direct query used to obtain the suggestion.
position	Suggestion position returned for that query.
source	google, bing, or youtube.
language	Requested language code.
country	Requested country code.
detectedAt	ISO timestamp for when the keyword was collected.

Results are deduplicated per source before being written.

---

Free run summary

Every run writes one free:

RUN_SUMMARY

record.

This means even a run where no keyword results are produced still creates a dataset record showing that the run completed and what happened.

The summary includes information such as:

• statusCounts
• keywordsWritten
• duplicatesDiscarded
• per-seed × source statuses

The RUN_SUMMARY record is not billed as a keyword result and does not count toward maxResults.

---

Pay only for written keyword results

This Actor uses Pay Per Event pricing.

You are charged for keyword records actually written to the dataset.

You are not charged as keyword results for:

• duplicate suggestions that are discarded
• filtered variants
• suggestions beyond maxResults
• the free RUN_SUMMARY record

The Pricing tab on this Actor page is always the authoritative source for current pricing and any run-start fee.

Cost-control options

The main controls affecting run size are:

seeds
mode
sources
modifiers
maxResults

For lower request volume, use:

{
  "mode": "suggest"
}

For larger long-tail harvesting, use:

{
  "mode": "expand"
}

---

Language and country targeting

Set the language and country in the input:

{
  "language": "pt",
  "country": "br"
}

This lets you run separate research for different markets instead of treating one global suggestion set as universal.

Example research workflow:

en + us
pt + br
de + de

Run each market independently and compare the resulting keyword datasets downstream.

---

Scheduling

This Actor is designed to work well with Apify Schedules.

Autocomplete suggestions can change over time, so recurring runs can help you observe how the long-tail around your topics evolves.

Recommended setup

1. Save your seed list and settings as an Apify Task.
2. Go to Console → Schedules → Create schedule.
3. Add the Task.
4. Choose your preferred recurring cadence.
5. Send the dataset downstream through Apify integrations or the API.

A weekly schedule is a practical starting point for many keyword-research workflows.

Example weekly cron:

0 6 * * 1

The Actor runs in the cloud, so your local computer does not need to remain on.

---

Example recurring SEO workflow

Input:

{
  "seeds": [
    "crm software",
    "sales automation",
    "email marketing"
  ],
  "mode": "expand",
  "sources": ["google", "bing"],
  "language": "en",
  "country": "us",
  "maxResults": 5000
}

Schedule it weekly.

Then send the output into:

• Google Sheets
• your database
• a webhook
• a content-research pipeline
• an SEO workflow
• a reporting system

Each run produces structured keyword records plus the free run summary.

---

Source resilience and run health

Autocomplete routes are public but undocumented, so response formats can change.

The Actor is designed to report source problems instead of silently emitting malformed keyword data.

Examples of controlled source/run conditions include:

API_CONTRACT_CHANGED
SOURCE_BLOCKED

If a source changes its response structure, the Actor reports the contract issue.

If a source returns a blocking response such as a redirect or 403, the Actor reports:

SOURCE_BLOCKED

rather than attempting to bypass the restriction.

When multiple sources are selected, a failure affecting one source does not automatically prevent healthy selected sources from continuing.

---

Operations and diagnostics

The run key-value store includes:

STATS

with operational information such as:

• per-seed × source summaries
• HTTP counters
• deduplication counts
• field-completeness health checks
• warnings

On failures, diagnostic information is written to:

ERRORS

The Actor is paced deliberately through requestDelayMs.

The minimum allowed delay is:

100 ms

and the default is:

250 ms

---

Important limits

Autocomplete is not search volume

Autocomplete suggestions are useful keyword and demand signals, but they are not search-volume measurements.

This Actor does not fabricate volume estimates.

If you need metrics such as:

• monthly search volume
• CPC
• keyword difficulty
• traffic estimates

combine this dataset with a separate source that provides those metrics.

Autocomplete depth comes from expansion

A single autocomplete request generally returns a limited suggestion list.

The Actor creates depth through:

• A–Z expansion
• digit expansion
• custom modifiers
• multiple seeds
• multiple sources
• market localization

Public routes can change

Google and Bing autocomplete routes used by the Actor are public but undocumented.

Their response formats can change without notice.

The Actor detects incompatible responses and reports them instead of deliberately converting unexpected data into keyword results.

Blocking is not bypassed

If a source blocks access through a response such as a redirect or 403, it becomes a controlled:

SOURCE_BLOCKED

condition.

The Actor does not attempt to bypass that restriction.

Character encoding

Some non-English autocomplete responses may use a non-UTF-8 charset.

The Actor decodes responses according to the returned charset so accented characters are preserved correctly.

For example:

imobiliária

should remain correctly encoded rather than becoming corrupted text.

---

Compliance notes

The Actor accesses public autocomplete routes without login credentials.

It does not require:

• Google API keys
• Bing API keys
• user accounts
• cookies from your account
• browser automation
• an LLM

Public availability does not imply unrestricted permission.

Provider terms still apply, and public autocomplete routes can change or become unavailable.

The Actor uses deliberate request pacing and reports blocking rather than attempting to evade it.

---

FAQ

Do I need a Google API key?

No.

The Actor accesses public autocomplete endpoints and does not require a Google API key.

Do I need a Bing API key?

No.

The Actor accesses the supported public autocomplete route without requiring a Bing API key.

Does it require a Google, Microsoft, or YouTube account?

No.

No login is required.

Does it use browser automation?

No.

The Actor makes direct requests to the supported autocomplete routes.

Does it use an LLM to generate keywords?

No.

Keyword records come from autocomplete responses rather than LLM-generated suggestions.

Which sources are supported?

The Actor supports:

google
bing
youtube

What is the difference between expand and suggest?

expand generates the seed plus A–Z and 0–9 query variants, along with configured modifiers.

suggest sends only the direct seed query.

Use expand for broad long-tail harvesting and suggest for focused direct autocomplete checks.

How many seeds can I use?

Up to 100 seed keywords per run.

Can I target Brazil or another market?

Yes.

Example:

{
  "language": "pt",
  "country": "br"
}

Can I add my own keyword modifiers?

Yes.

Example:

{
  "modifiers": ["best", "how", "vs"]
}

Up to 20 modifiers are supported and are applied as prefixes and suffixes.

Does it return search volume?

No.

Autocomplete does not provide trustworthy public search-volume data through these routes.

Use a separate volume provider if you need volume estimates.

Are duplicate keywords billed?

Discarded duplicates are not billed as keyword results.

What happens when the maximum result count is reached?

Keywords beyond maxResults are not written as keyword results and therefore are not billed as keyword events.

What is RUN_SUMMARY?

It is one free summary record written on every run.

It provides run-level coverage and status information even when no keyword records are returned.

Can I schedule the Actor?

Yes.

Save the input as an Apify Task and run it through Apify Schedules.

Is this affiliated with Google, YouTube, Bing, or Microsoft?

No.

This is an unofficial community Actor and is not affiliated with, sponsored by, or endorsed by Google, YouTube, or Microsoft.

---

Part of the JM Forge suite

Also from the same developer:

• Google Maps Business Scraper — Extract public Google Maps business leads with names, categories, addresses, phones, websites, and ratings.

• Amazon Product Search Scraper — Extract Amazon product listings, prices, discounts, ratings, reviews, ASINs, and images across seven marketplaces.

• Amazon Best Sellers Scraper — Extract Amazon Best Sellers rankings, ASINs, prices, ratings, and reviews from US and UK category lists.

Support

Report issues in the Issues tab of this actor — I respond within 24h. Feature requests welcome.