YouTube

View API reference · Tag YouTube

We built YouTube coverage so a founder interview, a conference talk, or a product review becomes structured data instead of something you have to sit and watch. Search finds the inventory; transcripts turn an hour of video into text you can read in seconds.

This is similar to a YouTube channel page, a video's description and chapter list, or the comment section — as fields you can query.

Inputs and pagination

Every list route returns nextPageToken. Send it back in the nextPageToken field on the next call; null means there are no more pages.

Durations come back as durationSeconds, already a number. Do not parse "42:18" strings.

Shorts work like any other video. Search results list Shorts separately (shorts), and the transcript route takes a Shorts link (youtube.com/shorts/…) as well as watch, youtu.be, embed, and live links.

What you can call

Example — POST /v1/youtube/transcript:

{
  "apiKey": "YOUR_API_KEY",
  "videoId": "xB4Tc2nKQvE"
}

Using it effectively

Reach for the transcript before the comments. If your question is "what did they say about X", a transcript lookup answers it in one call. Comments are a separate, paginated, per-page cost and they answer a different question — who cared enough to respond.

Let chapters do the navigation. The video details response returns chapters and keyMoments with timestamps. Match those against startSeconds on the transcript segments to jump straight to the pricing discussion or the competitor mention without reading the whole thing.

Search to discover, details to commit. Video search is how you build a candidate list cheaply. Only spend video details and transcript calls on the videos that survive a title-and-view-count filter.

Check the language before requesting one. Video details tells you which transcript languages exist. Asking for a transcript in a language the video does not have wastes a call.

Watch the tighter search limits. Channel and search routes allow 60 requests per minute, where video details, transcripts, and comments allow 300. Discovery is the part of your job that needs pacing.

Use cases

Turn a prospect's podcast appearance into an outbound hook. Run a video search for their name plus "podcast" or "interview", pick the recent hit, then pull the transcript on its videoId. Quoting something they actually said beats any generic opener, and it takes two calls.

Map the creators in a category. Search your category keyword, collect the distinct channels off the results, then run a channel lookup on each channelIdentifier to get subscriber counts and page their catalog. Rank by subscribers against upload frequency to find who is both large and active.

Mine competitor product reviews. Search the competitor's product name, call video details on the highest-view results and read chapters for the criticism sections, then pull the transcript for the exact wording. That is your objection-handling material.

Find people already engaged with a topic. Pull the comments on a video squarely in your category. The commenters have self-selected into caring. Note that reaching them needs a name or LinkedIn URL — a YouTube display name is not a lookup input.

Credits

Every YouTube call costs 2 credits — per request on details and transcripts, per page of results on channel, search, and comments. Search and channel routes allow 60 requests per minute; details, transcript, and comments allow 300 per minute. Back off on HTTP 429. chargeInfo on the response is authoritative — see Billing.

A video with no transcript returns 404 ("Transcript not found") and is still charged, because the lookup ran. If YouTube itself is unavailable, the call fails and isn't charged; retry shortly.

Credit amounts are defaults — custom pricing is available for bulk use cases.

Related: Social media APIs · TikTok transcripts · X (Twitter)