YouTube Autocomplete Keyword Scraper: a-z, questions, CSV/JSON

YouTube autocomplete as CSV/JSON rows. Give it seed keywords; it requests suggestions for the seed, seed + a to z, question words, and prepositions (comparisons and numbers only if you enable those expansions), for the language you set. One flat row per unique suggestion per seed, each with a stable id.

What you get

Field	Example
id	stable per seed, language, query and suggestion
keyword	coffee grinder (your seed)
query	coffee grinder b (the exact string sent)
suggestion	coffee grinder burr
rank	1 (position in the engine's list for that query)
expansion	direct, alphabet, question, preposition, comparison, number, depth
modifier	b, how, for, vs
depth	1 or 2
language, country	en, us
engine	youtube
fetchedAt	ISO timestamp

Rows are deduplicated per seed (case-insensitive), so a suggestion that appears under several modifiers is counted once, under the first query that produced it.

Where this sits on the shelf

Apify Store search for "youtube autocomplete", top 10 results on 2026-09-11, read from the Store API. Users, runs and the failed-or-timed-out share are the Store's 30-day public stats for each listing: an Apify Store snapshot, not a live quote.

Listing	Users (30d)	Runs (30d)	Failed or timed out
sian.agency/youtube-auto-complete-and-query-suggestion	13	64	3%
johnatan029/keyword-trend-monitor	1	15	0%
maged120/youtube-music-autocomplete	0	29	0%
aitorsm/keyword-fanout	17	68	0%
lofomachines/keyword-shitter-autocomplete-scraper	0	31	0%
karamelo/youtube-keywords-shitter	10	96	0%
solidcode/youtube-keyword-scraper	4	37	0%
ninhothedev/youtube-keyword-scraper	4	72	0%
maximedupre/youtube-keywords-discovery-tool	3	63	0%
scrapemint/google-keyword-suggestions-scraper	2	43	0%
tallyrake/youtube-autocomplete-scraper (this actor)	0	1	0 of 1

Same engine, other search boxes: TikTok search suggestions, Google, Amazon, Bing, DuckDuckGo. Same input shape, same flat rows, one run per engine.

Recipes

Content brief for one topic. Seeds: your topic. Expansions: questions, prepositions. Depth 1. Output: the questions and phrasings YouTube suggests for the topic, each with the engine rank on the query that first produced it (dedup keeps the first hit, not a single global order).

Keyword list for a niche. Seeds: 10 to 50 head terms. Expansions: alphabet. Depth 2. YouTube returns at most 10 suggestions per query, and maxQueriesPerKeyword (default 300) caps the requests per seed, so one seed can produce at most 3,000 dataset rows before dedup. The output is unique dataset rows, not queries. For 10 to 50 seeds there is no measured run yet; start at concurrency 3 without a proxy and turn on Apify Proxy only if the log shows 429 or 403 responses.

Localized suggestions. Same seeds, language set per run. Rows carry the code so runs can be merged.

Brand and competitor monitoring. Seeds: brand names. Expansions: comparisons, questions. Schedule daily. Diff the suggestion column between runs.

Input

Paste seeds one per line. The default expansions send 52 queries per seed (1 seed + 26 letters + 14 question words + 11 prepositions = 52). Comparisons and numbers are not in that 52; they are extra expansions. One seed sends many queries; the output is unique dataset rows, not queries. See the measured yield below for what 52 queries produced.

Reliability and cost

Measured on 2026-09-11 on Apify: 2 seeds (coffee grinder, pytrends 429), expansions alphabet + questions + prepositions (52 queries per seed), concurrency 3, no proxy, en/us: 104 queries produced 333 dataset rows (about 3.2 per query), 0 rate-limited responses, 13 seconds. That run used $0.0020 of platform compute. Per seed in that run: coffee grinder 323 rows, pytrends 429 10 rows.

Billing terms are on the Pricing tab of the listing. Apify platform usage is a separate charge and is not the listing price. Empty queries produce no rows. Every request is paced with a jittered delay (pacingMs, default 200 ms). If the engine answers 429 or 403, the run rotates its session and retries; that path was not exercised in the measured run.

Cost limit

If a run reaches the "Maximum total charge per run" you set, it stops in process rather than being aborted by the platform: the run ends with status SUCCEEDED, and the run status message on the run detail says how many rows were written and how many were found but not written. Re-running the same input writes the same prefix again, so raise the cap before re-running if you want the rest. Measured on this actor on 2026-09-14: a cap that stopped at 40 written rows on a three-seed, all-expansions, depth-2 input reported 1,222 found but not written; a cap that stopped at 200 written rows on the one-seed default input (coffee grinder, 52 queries) reported 123 found but not written.

Bigger jobs

Need this as a one-off file instead of a run? Ask for a quote at https://tallyrake.com/datasets/. We may decline. A quote binds nothing, and a job exists only when we confirm it in writing, before any payment.

Limits

This is YouTube's suggestion list for the language you set, at the time of the run. It carries no search volume; pair it with a Google Trends actor for demand. YouTube returns at most 10 suggestions per query. Export CSV for Excel and Google Sheets. Export JSON for pipelines.

Reviews

If you have run this Actor, a review on its Store page helps the next person decide, and a low score with a reason is as welcome as a high one. Write it in your own words. Nothing is offered in return, and a review changes nothing about the support you get.