# How to Find YouTube Channels by Topic

To find YouTube channels by topic, search for relevant videos, check which channels cover the topic often, and keep one entry per confirmed channel. Start with a manual review of public videos and channel pages. Use a stable channel ID to combine repeated results. Then check any subscriber rule, such as at least 20,000 subscribers. One matching video helps you find a channel. It does not prove that the channel focuses on that topic.

For a small list, you can do this work in a spreadsheet. A collection tool can help you save video links and channel details. You must still review the content and decide whether each channel meets your rules.

This guide builds a channel directory: a list of channels with source links, topic evidence, public profile checks, and clear decisions. It does not find every matching channel on YouTube. It also does not judge channel quality, predict growth, or identify the people who watch a channel.

## 1. Define what a matching channel must show

Write the purpose of the list before you search. For example, you might need channels that teach beginner ceramic glazing. That is narrower than channels that mention ceramics.

Separate the topic rule from the subscriber rule. A channel can meet one rule and fail the other.

Use three decision states for each rule:

- Accepted: The checked evidence meets the rule.
- Excluded: The checked evidence fails the rule.
- Unresolved: The evidence cannot support a decision.

These are labels you create in your research sheet. They are not documented collection-tool output values.

Write down what you will inspect. A proposed rule could require at least four relevant instructional videos among five checked recent uploads. Choose a rule that fits your purpose. This sample rule is a research choice, not a tested standard for channel discovery.

Define “relevant” too. A video that teaches a glazing step might qualify. A studio tour that briefly shows glazed pottery might not. Save a short reason for each decision.

Your subscriber rule also needs precise wording. “At least 20,000 subscribers” means the evidence must support a count of 20,000 or more. A rounded display may leave that decision unresolved.

## 2. Search manually and save the source videos

Start with several plain phrases that describe the topic and the viewer's task. For the glazing example, proposed phrases could include beginner ceramic glazing tutorial, pottery glaze application lesson, and ceramic glazing mistakes for beginners.

These phrases are search inputs. They are not evidence of search volume or demand.

For each search:

1. Record the exact phrase and the time you searched.
2. Set a limit on how many candidate videos you will inspect.
3. Record the sort setting if it is available.
4. Open each candidate and check the content needed for your topic decision.
5. Save the video link, channel link, and a short relevance note.
6. Keep uncertain candidates as unresolved.

Do not assign a topic from the title alone. A title can help you choose what to inspect. It cannot replace that inspection. If you cannot review the video, record that limit.

Treat Shorts and longer videos according to your stated purpose. A short demonstration may support a narrow skill. Your list may require fuller lessons. State that difference before you accept or reject a channel.

Stop at the boundary you set. A limited search gives you candidates from those phrases and observations. It does not show all channels on the topic. Nor does a channel's absence from your results prove that it does not cover the topic.

## 3. Use the Search Scraper when you need saved records

An Actor is a hosted data collection program. Chronometrica's YouTube Search Scraper collects discovered public videos and Shorts. Its documented fields include video identity, channel identity, query context, and available public metrics. A query is the search phrase sent to the collector. [Search Scraper public contract, checked 25 October 2026; no paid run](https://api.apify.com/v2/acts/Tv5YxlFwQjqWpwlLR/builds/default)

This fits the first collection step. It does not search an entire channel population and filter it by topic or subscriber count. Apply the channel rules after discovery and review.

The following small input matches the checked public input schema. A schema defines accepted fields and value types. This input was checked against that schema. It was not run.

```json
{
  "enableQueryExpansion": false,
  "includeShorts": true,
  "maxResults": 10,
  "queries": [
    "beginner ceramic glazing tutorial"
  ],
  "searchDistribution": "balanced"
}
```

3. Use the Search Scraper when you need saved records

Replace the sample phrase with your topic. Keep the sample if you are researching beginner ceramic glazing. Do not assume the example phrase suits another subject.

The settings have these documented meanings:

- `queries` contains the search phrases.
- `maxResults` limits saved unique video or Short candidates across the run. Here, the limit is 10 videos, not 10 channels.
- `enableQueryExpansion` set to `false` prevents the collector from adding YouTube autocomplete suggestions.
- `includeShorts` set to `true` allows Shorts-style cards when public search data exposes them.
- `searchDistribution` set to `balanced` rotates across queries until the overall limit or source and page limits are reached.

With one query, there is no second query to share the collection limit. If you later add phrases, keep the overall limit in view. Do not interpret it as a separate allowance for every phrase. [Search Scraper input rules, checked 25 October 2026; documented behavior](https://api.apify.com/v2/acts/Tv5YxlFwQjqWpwlLR/builds/default)

Inspect the other saved settings before you start. The sample leaves some controls unspecified. The checked schema gives `enrichWithVideoMetrics` a default of `true`. This setting requests a second collection stage for richer public metrics. Collecting those metrics does not remove the need to inspect topic relevance.

For more detail about collection limits, see the [bounded YouTube search guide](/en/articles/chronometrica-vs-streamers-youtube-search/). Use it to set the collection boundary while keeping the goal here focused on channels.

The public [YouTube Video Topic Discovery preset, presence checked 25 October 2026; saved input not verified](https://apify.com/chronometrica/youtube-search-scraper/examples/youtube-video-topic-discovery) provides another starting point. A preset is saved input for an Actor. Inspect its settings and replace its sample values before use. Public presence does not prove that the preset was run or that its saved input matches this example.

## 4. Keep a video evidence table

Keep the original video records before you combine them into channels. This lets another reader trace each channel decision to the videos you inspected.

Save these documented fields when they are available:

| Field | Reason to keep it |
| --- | --- |
| `postId` and `postUrl` | Identify the source video and let you return to it |
| `channelId` and `channelUrl` | Connect the video to a channel |
| `channelName` and `channelHandle` | Show readable channel details |
| `rootQuery`, `query`, and `querySource` | Retain the search context |
| `pageIndex` and `pageRank` | Retain the returned search position |
| `postType` and `isShort` | Retain the returned format information |
| `publishedAt` | Keep the available publication date |
| `status`, `statusReason`, and `observedAt` | Keep the collection outcome and observation time |

These fields describe collected records. Search position does not prove topic fit or channel quality. A missing publication date stays unknown. Do not make a video recent because it appeared near the top of the results. [Search Scraper output contract, checked 25 October 2026; no runtime test](https://api.apify.com/v2/acts/Tv5YxlFwQjqWpwlLR/builds/default)

Add your own review columns beside the collected fields. Include the relevance state, a short reason, the part you inspected, and the review time. Mark these as reader-created columns so nobody mistakes them for automatic topic analysis.

Keep missing values missing. A blank subscriber count does not mean zero subscribers. An unreviewed video does not mean an irrelevant video.

## 5. Combine video evidence by stable channel ID

A channel ID is an identifier for a channel. Use `channelId` as the key for grouping records. Grouping means putting video records with the same channel ID together.

A channel name is not a stable identity key. Do not merge records because their names look alike. If the channel ID is absent, retain the source record and mark channel identity unresolved.

For each confirmed channel ID:

1. Create one channel entry.
2. Keep the set of source video IDs and links.
3. Count each distinct evidence video once.
4. Preserve the search context that led to the channel.
5. Keep any identity problem visible.

This removes duplicate channel entries without losing the supporting evidence. It also keeps the units clear. Six video records can identify three channels. That does not establish six creators, or even three distinct people. A channel is the unit of this directory.

The collector's saved-video count and your final channel count answer different questions. Report both when they help explain the work. Do not call a video limit a channel limit.

## 6. Check the channel's recent content and public facts

Open each channel page and inspect the recent-upload sample you defined. Record the sample links and what you checked. Confirm that you have enough date or page evidence to call the sample recent.

Keep the discovery evidence separate from this channel review. A source video can be relevant while the channel's wider focus remains unresolved. Several matching search results can also come from one series within a mixed-topic channel.

Describe only what your sample supports. “Four of five checked recent uploads teach beginner glazing” is more precise than “This channel only covers glazing.” A small sample does not establish the subject of every upload.

For known-channel profile facts, Chronometrica's YouTube Channel Scraper accepts public channel URLs and handles through `startUrls`. Its documented output includes `channelId`, `channelUrl`, `subscribersCount`, `subscribersText`, `countPrecision`, `status`, `statusReason`, and `observedAt`. [Channel Scraper public contract, checked 25 October 2026; no paid run](https://api.apify.com/v2/acts/HjtSheUDKuElb7oxP/builds/default)

Use that collection step after you have candidate channels. It does not replace the recent-content review. The [channel profile table guide](/en/articles/chronometrica-vs-streamers-youtube-channel-profiles/) explains how to retain identity, precision, and unresolved profile attempts.

These are public records. They do not supply owner-only YouTube Studio records, private audience demographics, or a list of viewers. A public subscriber count cannot show whether two channels reach the same people.

## 7. Apply the subscriber rule without losing uncertainty

Keep the subscriber number with its displayed text, precision information, outcome, and check time. Precision describes how exact the reported count is.

For a strict minimum of 20,000 subscribers, use this rule:

| Available evidence | Subscriber decision |
| --- | --- |
| Usable exact count of at least 20,000 | Accepted |
| Usable exact count below 20,000 | Excluded |
| Rounded display of 20K without enough precision to decide | Unresolved |
| Missing or unusable count | Unresolved |

A numeric field alone does not prove exactness. Read the precision and outcome fields before you compare the value with the minimum. Preserve the displayed text too. [Channel Scraper count and outcome fields, checked 25 October 2026; documented contract](https://api.apify.com/v2/acts/HjtSheUDKuElb7oxP/builds/default)

Do not silently change “at least 20,000” into “displays about 20K.” Those are different rules. If an approximate size band suits your purpose better, record that new rule and assess the channels under it.

Keep topic and subscriber decisions in separate columns. For a list that requires both rules, accept a channel only when both are accepted. Exclude it when either rule clearly fails. Otherwise, leave the final decision unresolved.

## 8. Work through an original fictional example

The following example is original fiction. It is not Actor output, a completed run, or evidence about real channels.

Assume six distinct, reviewed source videos fit beginner ceramic glazing. Three belong to channel A, two to channel B, and one to channel C. All three channels have confirmed fictional channel IDs.

The arithmetic is:

`3 videos + 2 videos + 1 video = 6 relevant videos`

`1 channel A + 1 channel B + 1 channel C = 3 unique channels`

The list requires at least four relevant lessons among five checked recent uploads. It also requires at least 20,000 subscribers. These are assumed rules for this example.

| Channel | Discovery evidence | Recent-content review | Subscriber evidence | Final decision |
| --- | --- | --- | --- | --- |
| A | 3 relevant source videos | 4 of 5 checked uploads fit | Rounded 20K display | Unresolved: subscriber minimum |
| B | 2 relevant source videos | 2 of 5 fit; one relevant series in a mixed-topic sample | Assumed exact count of 24,600 | Excluded: topic rule |
| C | 1 relevant source video | Recent sample not reviewed | Count missing | Unresolved: topic and subscribers |

Channel A meets the fictional topic rule. Its reviewed share is:

`4 relevant uploads / 5 reviewed uploads × 100 = 80% of reviewed uploads`

That percentage describes the five-video sample. It does not describe every upload or the channel's audience. The rounded 20K display leaves the strict subscriber minimum unresolved.

Channel B meets the subscriber rule under the fictional exact-count assumption:

`24,600 subscribers − 20,000 subscribers = 4,600 subscribers above the minimum`

It still fails the topic rule. Channel C has a relevant discovery video, but its recent channel sample remains unreviewed. Its missing subscriber count stays missing.

The result is zero fully accepted channels, one excluded channel, and two unresolved channels. Zero accepted channels is a decision count. It does not mean that any channel has zero subscribers.

## 9. Export the decisions and check the list

Keep accepted, excluded, and unresolved entries in the saved directory. You can make a separate view of accepted channels for later work. Keep the full record so missing facts and exclusions remain visible.

Your channel table should contain these groups of fields:

| Field group | Save these details |
| --- | --- |
| Identity | Channel ID, channel URL, name, and identity state |
| Discovery | Source video IDs, video links, and search phrases |
| Topic review | Rule, sample links, review time, decision, and reason |
| Subscriber check | Count, displayed text, precision, outcome, and check time |
| Final decision | Accepted, excluded, or unresolved, with a reason |
| Collection boundary | Search limits, collection time, and relevant settings |

Before you use the list, check that each confirmed channel appears once. Check that each topic decision has inspected evidence. Check that every subscriber decision respects precision. Confirm that unresolved records remain visible.

State the boundary beside the export: the phrases searched, the collection limit, the actual collection time, and the recent-upload sample rule. These facts explain what the list covers. They do not measure its share of all matching YouTube channels.

## Take the next action

Write one topic rule and one optional subscriber rule. Review a small manual sample first. If you need saved discovery records, open the [YouTube Search Scraper](https://apify.com/chronometrica/youtube-search-scraper), replace the sample query, and inspect the settings before starting. Any run you start is a new action; this guide reports no paid run.

Resolve the missing fact that blocks your next decision. Channel A in the fictional example needs better subscriber evidence. Channel C needs a recent-content review and a subscriber check. Refresh changed public facts before later outreach or commercial use.

If your next task is to compare selected videos, use the [YouTube reference video guide](/en/articles/youtube-reference-panel-design/) to define that separate selection. The channel directory gives you checked sources for that work.

## Sources

- [Public reader question about finding channels by topic; observed 12 August 2026; publication date unknown; one question does not establish demand](https://www.reddit.com/r/scrapingtheweb/comments/1vgxrz7/beginner_question_is_there_an_easy_way_to_find/)
- [YouTube Search Scraper public default build 1.0.34; checked 25 October 2026; static contract inspection, no paid runtime test](https://api.apify.com/v2/acts/Tv5YxlFwQjqWpwlLR/builds/default)
- [YouTube Channel Scraper public default build 0.1.21; checked 25 October 2026; static contract inspection, no paid runtime test](https://api.apify.com/v2/acts/HjtSheUDKuElb7oxP/builds/default)
- [YouTube Video Topic Discovery preset; public presence checked 25 October 2026; saved input not authenticated, no paid run](https://apify.com/chronometrica/youtube-search-scraper/examples/youtube-video-topic-discovery)