eBay Keyword Suggestions

What eBay's own search box completes your keywords to, on the eBay site you pick, as a dataset: one row per unique suggestion, with the seed it came from, the exact query that produced it, its position in eBay's list, and eBay's category hints for that query.

eBay completes a partial query from what buyers on that site actually type. That makes its suggestions the closest public thing to a demand signal eBay offers: a phrase eBay completes is a phrase buyers search, and a seed with no completions is a phrase they do not. This Actor calls the same endpoint the eBay search box calls, expands every seed letter by letter (seed a ... seed z, seed 0 ... seed 9) and, if you ask, goes one level deeper.

Use case

• Listing titles - put the words buyers type into your 80 characters, not the ones you would guess.
• Product research - see which variants, models and sizes buyers ask for ("lego t" -> "lego technic porsche 911 gt3 rs 42056").
• Cross-border selling - the same seed on ebay.com, ebay.co.uk and ebay.de completes differently ("lego todesstern" exists only on ebay.de).
• Killing ideas cheaply - a phrase eBay will not complete is a phrase nobody searches.

What you get

field	meaning
keyword	the suggestion, exactly as eBay returned it
marketplace	the eBay site asked (US, UK, DE, ...)
seed	the seed keyword it was reached from
query	the exact query sent to eBay that returned it
position	1-based position in that query's list (eBay's order)
depth	0 seed, 1 letter expansion, 2 a suggestion queried as its own seed
query_categories	eBay's category hints for that query, [{id, name}]
read_at	when the query was answered (UTC)

A run also writes RUN_SUMMARY to its key-value store: queries sent, empty answers, rows pushed, and what stopped the run if a cap did.

Sample output

Three rows from a real run (eBay UK, seed vintage lamp, 2026-10-31):

[
  {
    "keyword": "vintage lampshade",
    "marketplace": "UK",
    "seed": "vintage lamp",
    "query": "vintage lamp",
    "position": 1,
    "depth": 0,
    "query_categories": [
      {
        "id": "20697",
        "name": "Home Lighting"
      },
      {
        "id": "1404",
        "name": "Collectable Lighting & Lamps"
      }
    ],
    "read_at": "2026-10-31T19:08:51.870Z"
  },
  {
    "keyword": "vintage lamp",
    "marketplace": "UK",
    "seed": "vintage lamp",
    "query": "vintage lamp",
    "position": 2,
    "depth": 0,
    "query_categories": [
      {
        "id": "20697",
        "name": "Home Lighting"
      },
      {
        "id": "1404",
        "name": "Collectable Lighting & Lamps"
      }
    ],
    "read_at": "2026-10-31T19:08:51.870Z"
  },
  {
    "keyword": "vintage lamp for restoration",
    "marketplace": "UK",
    "seed": "vintage lamp",
    "query": "vintage lamp f",
    "position": 1,
    "depth": 1,
    "query_categories": [
      {
        "id": "1404",
        "name": "Collectable Lighting & Lamps"
      }
    ],
    "read_at": "2026-10-31T19:08:52.044Z"
  }
]

Sample input

{
  "keywords": ["vintage lamp", "lego technic"],
  "marketplace": "UK",
  "expand": true,
  "depth": 1,
  "maxSuggestions": 500
}

Input

• keywords - the seeds to expand. One query each, plus 36 more per seed when expanding.
• marketplace - which eBay site to ask: US, UK, DE, CA, AU, FR, IT, ES, AT, CH, IE, NL, BE, PL. Default US.
• expand - also query each seed followed by a space and each of a-z, 0-9. On by default. This is where the long tail comes from.
• depth - 0 seeds only; 1 seeds plus expansion; 2 also query every suggestion found as its own seed, one level deep.
• maxQueries, maxSuggestions - caps on requests sent and unique rows pushed. Defaults 2,000 and 10,000; 0 is unbounded. These caps are also your bill ceiling.
• concurrency, timeoutSecs - how the requests are paced.

Pricing

Pay per event, with no charge for starting the Actor and no charge for platform usage:

event	price	when
query	$0.002	one request sent to eBay's autosuggest
suggestion-row	$0.003	one unique suggestion pushed to the dataset

A suggestion seen again under another query is not charged again, and an empty answer costs one query and no rows. eBay answers at most 10 suggestions per query, so one seed expanded at depth 1 is 37 queries and at most 370 rows: at most $1.18. The default caps bound any run at $34.00; lower maxQueries and maxSuggestions to bound it below that.

Use it from code or an AI assistant

The API tab on this page has ready-made snippets for Node.js, Python and curl, including run-sync-get-dataset-items, which starts a run and returns the rows in one call. For Claude, ChatGPT or any MCP client, add https://mcp.apify.com?tools=vital_tuxedo/ebay-keyword-suggestions as a server and ask: what does eBay UK complete "vintage lamp" to?

FAQ

Does this report search volume? No. eBay publishes no search volumes, so there are none here. position is where the suggestion sat in one query's list - an ordinal, not a volume and not a difficulty score.

Does it sign in, or open listings? No. It calls one public endpoint, the one the eBay search box calls, and reads its answer. No cookie, no account, no listing page, no seller data, no scraping of search results.

Why did a seed return nothing? eBay completed it to nothing. That is a real answer about the phrase, and it costs one query event and no rows.

Do the suggestions change? Yes, with season and site, which is why every row carries read_at and marketplace.

More from this developer

• Etsy Keyword Suggestions - the same thing for Etsy's search box.
• Google Ads Transparency Scraper - every ad a competitor runs on Google, with copy, landing URLs, dates and countries.

Running locally

npm install
npm start

Input is read from storage/key_value_stores/default/INPUT.json; the dataset lands under storage/datasets/default/.

Built and maintained with Claude (an AI model by Anthropic).