YouTube Search Results Scraper: keyword search to CSV

Search YouTube by keyword and export video titles, watch URLs, channels, views and search positions. Use one query or a small keyword set to build a content-research dataset without downloading videos or supplying a YouTube API key.

Try it: open the saved Python and automation search example. Result events cost $1.50 per 1,000 exported videos on the Free plan, plus the Actor-start event. Duplicate videos across queries are exported once.

Run your first search

Paste this into the JSON input, run, then choose Dataset → Export → CSV:

{"queries": ["python tutorial"], "maxResults": 3}

The Actor inspects up to three candidate videos for that query. maxResults is per query, so two queries with a limit of three inspect at most six candidates. Duplicate removal or exclusions can produce fewer rows.

Real output

Selected fields from a September 10, 2026 run. Search order and view counts change:

{
  "video_id": "_uQrJ0TkZlc",
  "url": "https://www.youtube.com/watch?v=_uQrJ0TkZlc",
  "title": "Python Full Course for Beginners",
  "duration_seconds": 22447,
  "channel": "Programming with Mosh",
  "view_count": 48936637,
  "timestamp": null,
  "query": "python tutorial",
  "search_rank": 1
}

Fields	What they contain
video_id, url, title	Video identity and canonical watch URL
query, search_rank	First matching query and one-based position before exclusions or deduplication
channel, channel_id, channel_url, uploader_id	Channel metadata exposed by the search result
duration_seconds, view_count, channel_is_verified	Public result metrics when available
description, thumbnail, timestamp	Search-description snippet, image and optional timestamp; this is not a full video-detail lookup

timestamp and other optional fields can be null. The Actor does not infer missing dates, engagement or subscriber counts. Search positions describe this run's response, not a personalized or location-specific ranking guarantee.

Keyword monitoring without repeat exports

Input	Default	Limits
queries	["python tutorial"]	1–20 nonblank phrases, up to 300 characters each; exact duplicate phrases run once
maxResults	10	1–100 candidate videos per query
excludeVideoIds	[]	Up to 10,000 previously exported eleven-character IDs

To skip a known video, add its video_id:

{"queries": ["python tutorial"], "maxResults": 3, "excludeVideoIds": ["_uQrJ0TkZlc"]}

Exclusions do not make the Actor search beyond maxResults. The remaining rows keep their original search position. The Actor does not remember previous runs automatically; store your exported IDs and pass them into the next run.

When a video appears under several queries, the dataset keeps the first query only. If you need a separate ranking record for every keyword, run each query separately. This duplicate-removal behavior was added in September 2026.

Costs and run status

At Free-plan pricing, each exported video costs $0.0015. The Actor-start event costs $0.00005 per allocated GB, with a minimum of one event. At 512 MB, three exported videos cost $0.00455 in Actor events. Existing plan discounts remain available; the Pricing tab shows your applicable rate.

Excluded and duplicate videos do not produce result charges. Set a maximum charge in run options to limit spending. Source failures still incur the start event, and any valid rows already exported retain their result charges.

Open Output → Run summary for per-query returned/exported counts, duplicate and exclusion counts, failures and the stop reason. If any query fails, the run fails after saving its summary and any valid partial results. Inspect both before retrying, so already delivered videos can be excluded.

Export with Python

Install apify-client, set APIFY_TOKEN and APIFY_DATASET_ID, then export a completed run without starting another:

import csv
import os
from apify_client import ApifyClient

client = ApifyClient(os.environ["APIFY_TOKEN"])
rows = list(client.dataset(os.environ["APIFY_DATASET_ID"]).iterate_items())
fields = ["video_id", "title", "url", "channel", "view_count", "query", "search_rank"]
with open("youtube-search.csv", "w", newline="", encoding="utf-8") as output:
    writer = csv.DictWriter(output, fieldnames=fields, extrasaction="ignore")
    writer.writeheader()
    writer.writerows(rows)

Save separate Tasks for different research topics, schedule them in Apify and compare datasets by video_id. API, n8n and Make workflows can use the same saved input.

Choose the right YouTube tool

Need	Actor
Discover videos from keywords	This Actor
Export an existing playlist	YouTube Playlist Extractor
Read a known video's details	YouTube Video Metadata Scraper
Get available captions	YouTube Transcript Scraper

Common questions

Does it return every result? No. YouTube search is dynamic and this Actor caps each query at 100 candidates. A successful run is a bounded snapshot.

Can it filter search by country, language or upload date? These are not supported input controls. Do not treat missing timestamps as evidence that a video falls within a date range.

Does it download media or comments? No. It exports search metadata only.

Why are there fewer rows than the limit? The source may return fewer candidates, duplicate videos may overlap between queries, IDs may be excluded, or the charge limit may stop delivery. Inspect SUMMARY for the observed reason.