YouTube Question Backlog & FAQ Brief

Find the questions your audience keeps asking—and prepare your next useful answer.

Turn an existing YouTube comment export into a ranked reply queue and a readable FAQ brief. See recurring question topics, the original wording, source links when available, and whether a creator reply was observed in the supplied snapshot.

Built for creators, community managers, product marketers and agencies working with product tutorials, reviews and launch videos.

Try the sample

Leave Explore the sample FAQ brief on and click Start. The synthetic sample needs no key, login or collection step. Open Open the FAQ brief in the run outputs, or download the CSV reply queue.

For your own comments, turn the sample off and choose one source:

1. Select an existing Apify comment dataset in the input picker; or
2. Paste a JSON array of comment objects.

This version analyzes supplied comments. It does not scrape video URLs or collect live YouTube comments. Use your own export or collect comments with your preferred tool first. No AI API key is required.

What you get

Output	Purpose
HTML FAQ brief	Ranked topic cards, exact question evidence, response-coverage badges and the full reply queue
CSV reply queue	Questions, topics, likes, source links and observed creator-reply status
JSON report	Counts, grouped evidence, import statistics, method and warnings
Default dataset	One row per detected question candidate, ready for a workflow or spreadsheet

The brief suggests topics to cover. It does not invent answers or send replies.

Comment format

Minimum input:

{
  "comments": [
    {"id":"example-1","text":"Does this work with Android?","likes":12},
    {"id":"example-2","text":"Does this also work with Android?","likes":4},
    {"id":"example-3","text":"How much is the monthly plan?","likes":3}
  ]
}

These are synthetic examples. For clickable source links, include a valid videoId (11 characters) and the original comment id. Source links are generated from those IDs; arbitrary supplied URLs are not copied into reports.

Common aliases are accepted: text / comment / commentText / question, id / cid / commentId, likes / likeCount / voteCount, plus videoId or a recognized YouTube videoUrl. YouTube Data API thread objects containing snippet.topLevelComment are also accepted. Clearly marked flattened reply rows (type: "reply", replyToCid or parentId) are excluded and counted; they are not automatically joined to parents. Supply nested replies for coverage analysis. Exports with unsupported fields should be mapped to the documented format first.

You can re-import a question queue exported by this Actor. Its old computed reply status is not treated as source evidence: without original reply records and creator IDs, the new run correctly reports unknown coverage.

Creator replies and coverage

Optionally supply the creator's stable creatorChannelIds (24-character IDs starting UC, not handles). Add authorChannelId to comments and replies. Supply replies as objects in replies, with replyCount / totalReplyCount and repliesComplete where known.

• Creator reply observed: a supplied reply has an exact matching creator channel ID. This does not mean the question was answered well or fully.
• No creator reply observed: creator IDs are known and supplied reply records meet the tool's completeness checks. This describes the supplied snapshot only.
• Reply coverage unknown: creator identity, reply authors or reply completeness cannot be established. Missing replies are never treated as proof of an unanswered question.

For example, replyCount: 0 supplies a snapshot with no replies. A valid supplied reply array matching the declared count, or repliesComplete: true with valid reply records and no declared count, can establish supplied-record completeness. repliesComplete: false always means incomplete. Records with missing reply authors, duplicates or inconsistent counts stay conservative. These are validations of your source's claims, not an independent check against YouTube.

YouTube's thread response can omit some replies; do not label a partial export complete. See the official thread documentation.

How ranking and grouping work

The tool uses transparent English question patterns and conservative token overlap, without an LLM. It assigns Compatibility, Pricing, Setup, Availability, Features or Other topics. Similar wording may be grouped into an editorial topic; this is a heuristic, not a guarantee that questions have identical intent.

Questions with complete supplied coverage and no observed creator reply are prioritized. Unknown coverage remains visible. Repetition and likes help order the queue; likes are not buying intent or proof of demand. Check the original evidence before making a content decision. Sarcasm, rhetorical questions, context, paraphrases and unsupported languages can produce misses or false positives.

Limits, privacy and permissions

Up to 1,000 supplied comments and 8 MiB per import. Each comment text is bounded at 10,000 characters; replies at 500 per thread and 10,000 total. Dataset imports take a bounded insertion-order sample and report truncation. Counts apply to usable imported comments, never the entire audience.

The Actor uses Limited permissions. The selected source dataset grants READ only; it is never modified. It does not call collection Actors, request your personal Apify token, or contact social platforms. Reports retain supplied public question text and source IDs, while omitting author names, handles, profile images and unrelated extra fields. Review comment text before sharing reports; text itself may contain personal information. Manage storage retention in your Apify account.

Pricing

Free beta: no developer fee. Apify platform usage and storage costs apply. Any separate tool you use to collect comments has its own pricing. There is no hidden AI or collection fee in this Actor.

Troubleshooting

• No usable comments: check that objects contain a supported text field, and are viewer comments rather than creator-authored comments.
• No questions found: a valid outcome; inspect the supplied sample and English-language scope.
• Unknown reply coverage: add exact creator channel IDs and complete reply records if your source provides them. Do not guess completeness.
• Dataset access failed: select the dataset using the input picker, which grants the required READ permission.
• Interrupted delivery: the saved report remains available. The Actor does not blindly replay an ambiguous dataset push; check the delivery state before reusing the run.