Keyword Suggestions: Google, YouTube, Bing, Amazon Autocomplete

Turn one seed keyword into hundreds of long-tail keywords that people really type into a search box:

• Four search boxes: Google, YouTube, Bing and Amazon. Each keyword row says which one it came from.
• Expansion that finds the long tail: the seed followed by every letter and digit ("heat pump a", "heat pump b"…), question words ("how heat pump", "why heat pump"…) and comparison words ("heat pump vs", "heat pump cost"…).
• Depth 2: also ask for the suggestions of each suggestion.
• Any language and country, including the matching Amazon store (US, CA, MX, BR, UK, DE, FR, IT, ES, IN, JP, AU).
• No duplicates: each keyword appears once per seed and source, with what was typed to get it and its place in the suggestion list.

Why this one

• Fast and cheap. Each search-box request is about 1 KB, and no proxy is needed for most runs. If a site blocks the run, it switches to residential IPs by itself.
• Pairs with the Google Trends Scraper: find the keywords here, then check their trends there.

Input

Field	What it does
seeds	One keyword per line.
google, youtube, bing, amazon	Which search boxes to ask. Google is on by default.
alphabet, questions, compare	Which expansions to use.
depth	1 or 2.
maxKeywordsPerSeed	Max unique keywords for each seed from each source.
language, country	Such as en and us.

{
  "seeds": ["heat pump"],
  "google": true,
  "youtube": true,
  "amazon": true,
  "questions": true,
  "maxKeywordsPerSeed": 300
}

Output

One row per keyword: keyword, source, seed, typed (what was typed into the box), position, depth, words.

Real rows from a test run on 17 October 2026:

[
  {
    "keyword": "heat pump water heater",
    "source": "google",
    "seed": "heat pump",
    "typed": "heat pump",
    "modifier": null,
    "position": 2,
    "depth": 1,
    "words": 4
  },
  {
    "keyword": "9000 btu mini split heat pump",
    "source": "amazon",
    "seed": "heat pump",
    "typed": "heat pump",
    "modifier": null,
    "position": 2,
    "depth": 1,
    "words": 6
  }
]

Pricing

Pay per unique keyword returned. Errors are free.

Notes

Autocomplete shows what people search for, not how often. For search interest over time, use the Google Trends Scraper.