Views do not arrive from one place. The mix differs by platform and, more importantly, by video. Knowing which source produced a video changes what you should fix: a clip that lives on search and one that lives in a recommendation feed are edited differently, titled differently, and judged by different numbers.

Most platforms give you a traffic source breakdown somewhere in the analytics, though the labels differ and some categories get lumped together. The categories below are the ones worth separating mentally even when the dashboard merges them.

The six sources worth separating

The recommendation feed. TikTok's For You, the YouTube Shorts feed, the Reels feed and Explore. This is where most views on most accounts come from.

Search. Somebody typed something and your video answered it.

Profile and home tabs. People who already know you, opening the app and checking.

Shares and direct messages. The video arriving inside a private conversation.

Suggested and related. Your video placed beside a larger one because the topics are close.

External. Links from websites, newsletters, messaging apps and anywhere else off-platform.

Recommendation feeds judge the opening seconds

Feed distribution is decided by early signal. The platform shows a video to a small group, watches what they do in the first moments, and either widens the audience or stops. Nothing later in the video can rescue a weak opening, because most of the audience that would have seen it never will.

That makes the feed the source where the hook is the whole game, and where a good middle matters only for the audience you already earned. It also makes feed traffic volatile: a video can get a large push on a Tuesday and nothing on Thursday for reasons that were never visible to you.

Search is the only source that compounds

A search-driven video keeps earning after publication. It does not need to win an auction every time it is shown; it needs to be the best answer to a phrase people keep typing.

This is the source most small accounts ignore, and the one that pays back slowest and longest. A single well-targeted video can out-earn twenty feed hits over a year, because the feed hits stop the day the platform loses interest.

Profile visits are a decision point

When someone lands on your profile they are deciding one thing: whether the next video is worth their time. They will look at the first three posts they see and decide in a few seconds.

So treat the profile as a page, not a filing cabinet. Pin the three videos that best represent what you actually make. Make the bio say what the account is for in one line. If the pinned three are all the same off-topic video that went viral, the profile is actively misleading the people most likely to follow.

Shares are distribution you did not pay for

A share moves the video into a conversation between two people, which is a stronger endorsement than a like and reaches viewers with no ad-style resistance. It is also the one source where the growth is genuinely yours: the platform did not decide to show it, a person did.

Ask what makes somebody send a short to a friend. In practice it is one of two things: recognition ("this is you") or usefulness ("you need this"). Neither happens by accident. A video built around a specific, nameable situation gets sent; a video about "content strategy" gets watched and forgotten.

Suggested and related borrow another audience

The platform places your video next to a popular one when the topics are close enough. That is a reason to cover subjects that already have large videos beside them, and to use the vocabulary of those videos so the connection is obvious. A video about fixing audio in a specific app will be suggested beside other videos about that app; a video about "better sound" will not be connected to anything.

Reading your breakdown honestly

Two cautions. First, percentages on small accounts are noisy, and a single share can register as a large slice. Second, distinguish between a video that got eighty percent of its views from search, which will keep earning, and one that got eighty percent from the feed in a single push, which mostly will not.

The useful question is not "where did these views come from" but "which of these sources will still work next month without me doing anything".

Making one short findable in search

A checklist that takes two minutes:

Say the phrase a person would type, out loud, in the first sentence.
Put the same phrase on screen as text for the first few seconds.
Use it in the title and in the first line of the description.
Keep captions on. Auto-generated captions help both the viewer and the platform understand the audio.
Use two to four specific hashtags as topic labels, not a wall of generic ones.
Stay on one topic long enough for the platform to understand what the account is about.

That last point does more work than the first five. Accounts that publish everything from fishing to finance give the platform no reliable signal, so their videos get classified loosely and surfaced rarely.

When external traffic is worth the effort

Sending two hundred visitors to a short from a newsletter or a website feels productive and does not make the feed pick it up at all. External traffic is worth chasing when the destination is something you control, such as a page, a list or a product, and not when you are trying to inflate a video's count.

The first day is a distribution audition

Almost all feed traffic arrives in the first day or two, and what happens then decides whether more arrives. So the first day is not for celebrating or panicking. It is for reading three things: the swipe-away rate, the share count, and whether the video is reaching people outside your existing audience.

If the swipe-away rate in the first three seconds is high, the video is finished regardless of its quality, and reposting it will produce the same result. If shares are meaningful while views are modest, the video has the quality signal but not yet the distribution — and a follow-up on the same subject is worth making, because the platform has learned something about who wants it.

Diagnosing a video that died

The source breakdown separates three failures that otherwise look identical.

A video with a thousand views and almost no shares is a hook problem. A video with a thousand views, a decent share count and no search traffic is a topic problem: a good video with no demand behind it. A video with search traffic and low retention is answering the query badly — right subject, wrong answer.

Each points at a different edit, and without the breakdown every underperformer looks like the same failure. That is precisely how people conclude that short-form is random.

What a healthy mix looks like

On an account that has published consistently for months, a typical month is mostly feed traffic, with a persistent trickle from search, a small steady flow from profile visits, and enough sharing to produce the occasional spike.

The proportions matter less than the presence of all four. Feed-only months feel exciting and evaporate. The accounts that grow steadily always have a search and profile floor underneath the spikes.

The exception is a video that is genuinely a reference: an explainer people will link to from a forum or a client email. Those earn search and referral traffic together, which is the combination worth building toward.

Where Short-Form Views Actually Come From