YouTube Research Digest — YouTube transcript research and AI video summaries

Give this actor one topic keyword and it does the watching for you: it finds the top-ranking YouTube videos for that keyword, pulls their full transcripts, and returns a single structured research digest — ranked themes, points of consensus and disagreement, mentioned brands, and quotable moments — plus per-video notes. It is a YouTube transcript tool and a YouTube video summarizer in one call, built for competitor and market research instead of one-off caption downloads.

For content marketers, creators and research teams, one keyword returns one decision-ready digest of up to 20 videos for $1.30 or less — mapping a niche's video landscape in minutes instead of hours of footage. Output lands as JSON, CSV or Excel and plugs into n8n, Make, Zapier and AI agents (Claude, ChatGPT) through the Apify API and MCP.

What does YouTube Research Digest do?

You pass a keyword; the actor searches YouTube for the top videos, reads every transcript it can, extracts per-video notes (topics, brands, key claims, timestamped quotes), and synthesizes the whole corpus into one ranked digest — one dataset row per keyword, the digest plus the evidence it was built from.

What it does not do: it does not stream or download video files, it does not touch private or unlisted videos, and it does not need a Google or YouTube login. It reads the same public search results and transcripts a signed-out visitor can see.

Why use YouTube Research Digest

• One keyword, one digest. Up to 20 top videos condensed into a single ranked summary — no playlist to sit through.
• Full transcripts, not snippets. Every available transcript is read end to end, so the digest is grounded in what was actually said, with timestamps.
• Themes ranked, not just listed. Recurring topics are ranked by how many videos cover them, with explicit consensus and disagreement between creators.
• Brand and quote extraction. Every brand mention is surfaced with its contexts; quotable moments carry the videoId and startSec so you jump straight to the clip.
• Flat price per digest. $1.30 per keyword on the Free plan (down to $0.70 on Gold and above), no seat and no subscription. A keyword that returns zero videos is still a delivered notFound row.
• Pipeline-ready output. Structured JSON per row, exportable to CSV/Excel and callable from the API, MCP and no-code tools.

How to research a YouTube keyword

1. Open the Input tab and type a topic keyword under Keyword (for example standing desk).
2. Optionally set Country (two-letter ISO code, e.g. US) and Language (e.g. en) to target a specific YouTube market.
3. Optionally lower Number of videos (topN) if you want fewer than the default 20.
4. Click Start and wait a few minutes for the run to finish.
5. Open the Dataset tab and read the row: start with digest.themes, then drill into videos[] for the per-video notes and digest.quotableMoments for clip-ready lines.

Minimal input:

{
  "keyword": "standing desk"
}

Targeting a specific market:

{
  "keyword": "standing desk",
  "country": "US",
  "language": "en",
  "topN": 20
}

What data you get

One dataset row per keyword. Here is a real row, trimmed for readability:

{
  "keyword": "standing desk",
  "country": "US",
  "language": "en",
  "videosFound": 20,
  "videosAnalyzed": 18,
  "videos": [
    {
      "videoId": "aaa111",
      "title": "Best Standing Desks 2026 — Tested",
      "url": "https://www.youtube.com/watch?v=aaa111",
      "channel": "Desk Lab",
      "views": 482000,
      "position": 1,
      "topics": ["ergonomics", "motor noise", "stability"],
      "brands": ["Uplift", "Fully"],
      "keyClaims": ["Dual motors are quieter under load"],
      "quotes": [{ "text": "the Uplift barely wobbled at full height", "startSec": 42 }],
      "source": "primary"
    }
  ],
  "digest": {
    "themes": [
      { "theme": "Frame stability at full height", "rank": 1, "videoCount": 2 },
      { "theme": "Alternating sit/stand cadence", "rank": 2, "videoCount": 2 }
    ],
    "consensus": ["Dual-motor frames are more stable than single-motor"],
    "disagreements": ["Whether an anti-fatigue mat is necessary"],
    "brandMentions": [
      { "brand": "Uplift", "videoCount": 1, "contexts": ["praised for stability"] }
    ],
    "quotableMoments": [
      { "text": "the Uplift barely wobbled at full height", "videoId": "aaa111", "startSec": 42 }
    ],
    "source": "primary"
  },
  "notFound": false,
  "contentHash": "8ed50d…",
  "finishedAt": "2026-07-23T18:34:29Z"
}

Field by field:

Field	Type	What it means
keyword	string	The topic keyword you researched (echo of input)
country / language	string	The market used for search and transcripts (echo of input)
videosFound	int	How many videos the search returned
videosAnalyzed	int	How many of those had a usable transcript — coverage, not an error
videos[]	array	Per-video notes: videoId, title, url, channel, views, position, plus extracted topics, brands, keyClaims, quotes and a transcript source
digest.themes[]	array	Recurring themes, ranked, each with a videoCount
digest.consensus[] / digest.disagreements[]	array	Where creators agree and where they differ
digest.brandMentions[]	array	Each brand, how many videos mention it, and the contexts
digest.quotableMoments[]	array	Clip-ready lines with videoId and startSec
notFound	bool	true when the search returned zero videos — still a delivered row
contentHash	string	Deterministic hash of the transcript corpus; stable across runs of the same input
finishedAt	ISO datetime	When the digest was produced

Every row carries the same keys. A notFound row keeps the shape with an empty digest, so your parser never has to special-case it.

Input parameters

Field	Required	Default	Notes
keyword	yes	—	The topic keyword to research, e.g. standing desk
country	no	—	Two-letter ISO country code for the YouTube market, e.g. US, GB, DE
language	no	—	Two-letter language code for results and transcripts, e.g. en, de
topN	no	20	Number of top videos to analyze; maximum 20 in this version (one search page)

You can set these in the Input tab, or pass them as JSON through the API and the Apify MCP server (mcp.apify.com), which lets AI agents pick and run this actor from your prompt.

How much does a YouTube research digest cost?

Pay per event — no monthly seat, no subscription.

Event	Price (FREE tier)	When it fires
Actor start	$0.05, once per run	The run starts and validates your input
Research digest	$1.30 per keyword (Free plan)	Once per delivered digest row

Prices step down with your Apify Store tier:

	FREE	BRONZE	SILVER	GOLD+
Research digest	$1.30	$0.80	$0.75	$0.70

Everything in the digest — every theme, every brand mention, all per-video notes and quotable moments — is included in that single per-keyword price. A run needs a per-run budget of at least $1.35 (one start plus one digest) to begin.

You are never charged for the digest event when: a run stops at your plan's budget cap before doing any paid work, or a result is served from the 24-hour cache. Those cost the $0.05 start only.

You are charged for a keyword that returns notFound, because "no videos rank for this keyword" is a delivered, decision-ready answer.

YouTube Research Digest vs watching videos or generic summarizers

	This actor	Manual research	Single-video summarizer
Input	one keyword	search + watch, video by video	one video URL at a time
Scope	top 20 videos, synthesized	as many as you have hours for	one video
Themes across videos	ranked, with consensus/disagreement	in your head	none — single video only
Brand mentions	extracted across the corpus	manual note-taking	rarely
Quotable moments	with timestamps	manual scrubbing	sometimes
Output	JSON / CSV / API / MCP	notes	text blob
Price	from $0.70 per keyword digest	hours of your time	per-video or per-seat

A single-video summarizer answers "what is in this one video?" This actor answers "what is the whole niche saying, and where do creators agree or clash?" — the question behind real YouTube competitor research.

How to use it with AI agents and no-code tools

The output is shaped for software, not just for reading.

• n8n / Make / Zapier — trigger with a keyword, read the dataset, then branch on digest.themes or digest.brandMentions to fill a brief or a sheet.
• AI agents via Apify MCP — connect mcp.apify.com and ask an agent to research a keyword on YouTube. Because each row is structured, the agent reasons over it without you writing parsing logic.
• API — call the actor directly and fetch results as JSON or CSV:

curl -X POST "https://api.apify.com/v2/acts/doesaiknow~youtube-research-digest/runs?token=$APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{ "keyword": "standing desk", "country": "US" }'

What this actor does not do

• It does not download or store video or audio files — it reads transcripts only.
• It does not access private, unlisted, age-gated or members-only videos.
• It does not log in to Google or YouTube, and it holds no account credentials.
• It does not analyze more than 20 videos per run in this version; deeper pagination is planned.
• It does not guarantee every video has captions — videos without an available transcript lower videosAnalyzed but never fail the run.
• The digest is a research summary, not legal, medical or financial advice; verify claims before acting on them.

FAQ

Is this a YouTube transcript tool or a summarizer?

Both. It pulls full transcripts for the top videos on a keyword and then summarizes them into one ranked digest, so you get the raw transcript signal and the synthesized answer in a single run.

What happens if a video has no captions?

The row is still produced. That video's transcript source is null, it is excluded from the synthesis, and videosAnalyzed is lower than videosFound. Missing captions are coverage information, not an error.

What if no videos rank for my keyword?

You still get a row with notFound: true and an empty digest. That is a delivered answer — "nothing ranks here" is useful for research — and it is billed as one digest.

Can I research the same keyword in different countries?

Yes. Set country (and optionally language) per run. Each distinct market is a separate input, so standing desk in US and in GB are two different digests.

Are results cached?

Identical inputs within 24 hours are served from cache at no digest charge — you pay only the $0.05 start. Change any input field to force a fresh run.

Related actors

• Google Local Pack Tracker — track the Google local 3-pack and AI Overview per keyword and city.
• Amazon Listing Optimizer — audit and rewrite Amazon listings by ASIN with keyword gaps and a 0-100 score.
• AI Overview Disruption Monitor — watch how Google's AI Overview and AI Mode change who gets cited for your queries.

Legal & support

This actor reads publicly available YouTube data — the same search results and transcripts any visitor sees without logging in.

• No login, no account access, no gated content. It never signs in to Google or YouTube and never touches private, unlisted or members-only videos.
• No personal data. Output contains public video metadata, transcript-derived notes and a generated digest — no viewer identities, comments or contact details are collected.
• Public data only. Search results and transcripts are derived from public web content.
• A research aid, not advice. The digest summarizes what creators said; verify claims before you rely on them.

You are responsible for using the output in line with YouTube's terms and any laws that apply to you. Found a bug or need a new field? Open an issue on the actor's page — feature requests shape the roadmap. A review on the Store page helps other researchers find it.

SEO keywords

youtube research tool, youtube transcript tool, youtube video summarizer, youtube competitor research, video transcription api, youtube transcript api, youtube summary generator, summarize youtube videos, youtube keyword research, youtube content research, youtube market research, bulk youtube transcripts, youtube video analysis, youtube niche research, ai youtube summary, youtube themes extraction, youtube brand mentions, video research digest, youtube seo research, transcript extraction tool, youtube topic analysis, competitive video research, youtube content strategy, analyze youtube channel topics

Built by doesaiknow.