Bing Keywords Discovery Tool

Collect Bing autocomplete keywords from one or more seed terms. The Actor queries Bing's OpenSearch and autosuggest endpoints, merges their source-backed fields, removes duplicates, and returns a stable dataset for SEO and content research.

Input

Provide at least one value through keyword or keywords. When both are present, the Actor combines them and removes case-insensitive duplicates.

Field	Default	Purpose
keyword	content marketing in the UI	One seed keyword
keywords	[]	Up to 50 additional seeds
market	en-US	Bing language-country market
expandAlphabet	true	Query the seed followed by a-z
expandQuestions	false	Query common question words followed by the seed
expandNumbers	false	Query the seed followed by 0-9
maxKeywords	200	Maximum unique records (1-2000)
maxConcurrency	5	Expansion queries processed in parallel (1-10)
delayMs	150	Delay between query variants (0-5000 ms)
maxRequestRetries	2	Retries for timeouts, rate limits, and temporary 5xx failures
requestTimeoutSecs	15	Per-request timeout (5-60 seconds)
proxyConfiguration	proxy disabled	Apify or custom proxy configuration

Example:

{
  "keywords": ["machine learning", "data science"],
  "market": "en-US",
  "expandAlphabet": true,
  "expandQuestions": true,
  "expandNumbers": false,
  "maxKeywords": 200,
  "maxConcurrency": 5,
  "proxyConfiguration": { "useApifyProxy": false }
}

Dataset

Every record follows a typed 35-field contract. Important source fields include the keyword, Bing display text, Bing suggestion type, raw relevance score, Bing-provided suggestion URL, source query, seed, expansion, market, and contributing endpoints. Derived fields add a stable ID, seed position, normalized relevance, word and character counts, question/long-tail flags, seed relationship flags, a public Bing search URL, and consistent search metadata.

Optional attributes that Bing does not provide are omitted rather than emitted as null or fabricated. For example, an OpenSearch-only suggestion can legitimately have no autosuggest URL or type, and a base query has no expansion value. The checked-in samples/sample-record.json is from the live machine learning Bing suggestion response inspected in Chrome.

Deduplication uses a normalized keyword key, while each record's sources array shows whether osjson, as-html, or both contributed data. maxKeywords is enforced globally and allocated fairly across seeds. Query variants are interleaved and processed with bounded concurrency, so a large first seed cannot starve later seeds.

Reliability and security

• Validates HTTP status and content type before parsing.
• Safely rejects malformed or unexpected endpoint payloads.
• Retries only temporary failures with bounded backoff.
• Uses one stable configured proxy session per run, with an Apify-compatible session ID.
• Logs endpoint/status metadata without response bodies, credentials, or proxy URLs.
• Completes with an empty dataset for a valid no-result search.
• Fails the run when every Bing endpoint request fails.

Local verification

npm ci
npm test
apify validate-schema
apify run --purge --input-file INPUT.json
npm run validate:dataset
npm run audit:dataset

The audit checks the 35-field dataset contract, source URL shape, sensitive-key exclusion, input-schema hygiene, and at least 20 populated top-level fields per record. The checked-in tests cover minimal, multi-seed, question expansion, no-result, and invalid inputs.